SCHOOL CHOICE IN BANGLADESH - Institute of … · The motivation for the School Choice in...

December 2008

SCHOOL CHOICE IN BANGLADESH

Report to DFID Bangladesh

Naomi Hossain, Institute of Development Studies, University of Sussex

Heidi Tavakoli, Oxford Policy Management

School Choice in Bangladesh

i

Acknowledgements

The study was developed from an initial concept by Samer Al-Samarrai, and the team are grateful for his support throughout, as well as detailed comments on the methods and the analysis. The study was funded by the Department of International Development in Bangladesh, and the team benefited from the support and comments from the DFID Bangladesh team. Thanks are particularly due to Barbara Payne and Fazle Rabbani, and to participants in a discussion about the draft report in November 2008 for their comments. Valuable support is acknowledged from Sajidul Islam (UNICEF) and Manos Antoninis (OPM), as well as useful discussions with Masooda Bano (QEH, University of Oxford) and Stuart Cameron (University of Sussex, CREATE consortium).

The team was led by Naomi Hossain (team leader, IDS), Georgina Rawle (OPM) and Heidi Tavakoli (OPM). In Bangladesh, the team included qualitative and survey researchers and experts managed by Mitra and Associates: Abu Naser (education and participatory methods), Md Jahangir Sharif (operations, survey management), Mamunur Rashid and Md Bayazid Hasan (community research supervisors); Taznin Merin Lopa, Aloka Talukder, Razia Sultana Banhi and Nusrat Jahir Shukti (qualitative and community researchers); Omar Faruque Siddiki (qualitative, community and school research), Hedayet Hossain, Javed Iqbal, Yaqub Miah, Saiduzzaman and ATM Jorjis Alam (school researchers).

The views expressed in this report are those of the authors and not of DFID or the organisations of which they are members.


ii

Executive summary

This report presents findings from a study of School Choice in Bangladesh, conducted in four districts of Bangladesh in 2008. The objective of the study was to analyse household choices around basic education by exploring conditions of school supply; preferences and aspirations around education; and schooling decisions and experiences, including the economic, social and political outcomes from education. The research was designed to take into account the impacts of gender, poverty, economic opportunity, religiosity, and human security as factors likely to affect decisions around educational investment.

Background: the changing demand for education in Bangladesh

The motivation for the School Choice in Bangladesh study was to inform education policy in a context in which the demand for schooling has been changing: enrolment in registered and official schools at primary declined in the first half of the 2000s, while at secondary level, expansion stagnated. Boys‟ enrolments in the official system declined most, and they appear to have been moving to different – and new - types of institution. Declining quality in the official system at primary is likely to squeeze out two groups: the poorest, for whom the low returns from poor quality education mean investment in school may not be worthwhile, and the rich, who can afford better education, should it be available. Fewer poor boys move to NGO schools than girls, and it has been suggested that madrassahs of a range of types are likely to have been absorbing these groups in this changing context. The potential political implications, such as the possibility of rising Islamic militancy among a cohort of young men from poor backgrounds, have directed attention to the question of how school choice is changing.

Global debates on School Choice

The study is situated within global debates about school choice. These have been ideological debates focused on the impact of wider choice in creating market-like conditions of consumer choice in developed countries. Evidence on the impact of school choice has been inconclusive in terms of impacts on student achievement and school quality, and a case against policies of widening school choice includes evidence that wider school choice can increase inequality. In the Bangladesh context, the school choice framework helps to frame analysis of important education policy issues, including:

Diversification and expansion of education service provision

Accountability and governance in education services

Perceptions of quality in education provision, and

Equity in provision and outcomes.

Methods and approach

The study adopted a comparative case study methodology, involving four contrasting communities. Large-scale nationally representative survey data was not used, and the approach focused on enabling an in-depth understanding of household decision-making processes around education. The strategy adopted was multi-disciplinary, multi-method and multi-level, approaching the issues at individual, household, community, school and regional levels. National data was used to understand the overall context of school provision. Based on this four villages in four distinctly different districts were selected to represent


iii

characteristics of a) extreme poverty; b) high insecurity, including environmental insecurity; c); economic opportunity and high growth; and d) above average religiosity. Research instruments were developed, pre-tested, piloted and finalised to study individual, household, community, regional and school characteristics in each.

Findings

School supply

The findings support national data in illustrating that school supply has expanded and diversified rapidly. After the phase of expansion of registered and government schools in the 1990s, a second expansionary phase started in the 2000s, when NGO and kindergarten schools, new hybrid schools and a range of types of madrassah emerged. The vast majority of students remain within the general stream registered institutions at primary and secondary, but a significant minority attend religious schools, and a smaller group, mainly the rich and the poor, attend other types of school.

The diversity and magnitude of school supply varied widely by area in the present sample, suggesting supply is likely to vary across the country. The distance between home and school emerged as the most important influence over school choice. The issue of distance contains within it concerns about children‟s security; the extent of these concerns vary widely by area, and are most acute for adolescent girls. Sexual harassment and even the abduction and rape of one schoolgirl were reported, signalling the significance of security in shaping school choice.

The study findings do not support the view that wider school choice raises quality, for two reasons. First, common indicators of quality do not suggest that newer entrants into education service provision are of higher quality than existing actors: established schools remain better-endowed in terms of teacher qualifications and physical facilities. Second, schools in areas with more diverse school supply do not appear to perform better on standard attainment indicators than those in areas with fewer schools. However, standard indicators of quality may not necessarily closely reflect the features of schools that people take into account when selecting an institution. The lack of meaningful indicators appears to be most important with respect to measuring governance, management and parental participation in schools. The quantitative and qualitative evidence are somewhat contradictory: on paper, registered and government schools are better governed and managed. Qualitative evidence suggests, by contrast, that parents may experience more accountability and participation in the newer types of schools, such as KG schools or madrassahs.

Qualitative findings also suggest wider school choice may have led to some competition and improved performance in a few contexts, as well as evidence that wider choice may lead to increasing socioeconomic differentiation within education, as richer households withdraw from schools attended mainly by the poor. These findings are primarily qualitative insights, however, and more needs to be understood about how competition between schools takes place in practice, and at national trends in the educational choices of the middle classes. Evidence of widening education inequality gains some support from the wide variations found in spending on school.


iv

Preferences and aspirations

The study found significant normative change with respect to education. This included that a basic minimum of education was deemed necessary even among the poor. Livelihood and economic opportunities were important determinants of the strength of this norm and the level of education it implied as minimally necessary; this meant, again, that area effects were strong. There was also support for the idea of equivalence in investments in children by gender: up to primary, at least, there was very little evidence of gender differentiation in preferences and norms around education. Gender difference emerged at secondary, but more noticeably in the poorest areas, where early marriage and large dowry payments remained a crucial feature of household investment strategies around gender. Where international migration was a common livelihood option, the level of education deemed necessary for boys was that which was adequate to gain entry into that labour market; by contrast girls‟ education was seen in some contexts as enhancing marriage prospects. Madrassah and/or Arabic education were seen as particularly important for girls.

The desire for social status is an important factor shaping school preferences, and the social status considerations of local reference groups appears to have a significant impact on aspirations, again highlighting the importance of area effects. The widening of school choice appears to have been driven in part by demand among affluent and elite groups to mark distinctions between themselves and poorer households enrolling in government primary schools.

Parents appear to closely scrutinise and possess standards for teaching quality against which they judge school performance; although capacities to assess school performance varied, few seemed to entirely uninformed or apathetic about the quality of schools. Yet parents seem to lack opportunities or mechanisms through which to exact better performance when teaching quality falls short.

The apparent growth in preference for religious schooling appears to reflect a) the growth in supply of madrassah institutions of a wide range and type in recent years, many benefiting from private donations; b) change in official recognition and status of graduates of registered aliya madrassahs, now officially equivalent to general stream qualifications; c) the growth of demand for madrassah teachers accompanying the growth of the madrassah sub-sector itself; d) the impact of migration to the Middle East, which appears to increase demand for religious education; e) views on madrassah education as a source of religious blessings and social respect; and f) practical perspectives on aliya madrassahs usefully combining provision of religious with general education in a single site.

Schooling choices and outcomes

The educational opportunities of the present generation are considerably wider and deeper than was the case for their parents. They are more likely to enrol and to stay in school and to attend a wider range of types of school, particularly at primary, than their parents. However, over two-thirds in our sample still attend government primary schools, highlighting the significant limits to choice in practice. No significant gender differences could be identified in schooling choices - in itself a significant finding.

Nor could any particular patterns with respect to type of school chosen could be identified by poverty level; however, this is likely to reflect the sampling strategy and small sample size more than any real difference. Area effects appear to be strong, most notably in the levels of education households aspired to for their children.


v

Findings on the socioeconomic outcomes of schooling choice decisions were inconclusive, largely because the wider school choice available to households is relatively recent and the effects have yet to be fully played out in the labour market. One finding of note was that most men who were no longer in education reported their occupation as self-employment rather than agricultural activity, reflecting the growth of rural off-farm and non-farm economic activity. The level rather than type of education attained was the key determinant of women‟s labour force participation. Expectations of the level of education to be attained varied mainly by area, supporting the argument that area effects are strong influences of educational decisions, independent of poverty levels.

The study also explored outcomes in terms of socio-political attitudes of students emerging from madrassah compared to general stream schools. The findings are suggestive rather than definitive, of emergent attitudinal differences. These included an overall strong preference for democratically elected leadership, which seemed to be compatible with acceptance of other forms of rule (e.g. by experts or the military), but also a widely shared concern that democracy leads to raised levels of corruption. More significant differences emerged around citizen rights, in particular women‟s freedom to work, in which the views of madrassah students, particularly boys, were at distinct odds with those of general stream students, particularly girls.

Policy implications

The School Choice in Bangladesh study has documented social changes being brought about by changing attitudes and preferences, policies and market forces in basic education in four communities in Bangladesh. Given the homogeneity and centralisation of the basic education sector, other communities are likely to be experiencing many of the processes of change discussed here.

One finding with implications beyond the issue of school choice was that ease of access, incorporating parental concerns about the time, security, cost and effort involved in travelling distances to schools, remained a key consideration in education decision-making. A second finding was that accountability and participation are features of schools that parents consider in education decisions, but conventional indicators do not capture dimensions that parents appear to value.

With respect to school choice specifically, key conclusions to be considered with respect to policy development include that:

1. Wider choice may not lead to significant gains in terms of quality. This is partly because cost and access continue to limit effective choice. But it also reflects the challenges of improving accountability to parents, which may not be overcome simply by the existence of alternative schools.

2. Differences of quality appear to reflect differences of governance and management more than tangible indicators of physical facilities or teacher qualifications. This highlights the need for more meaningful indicators of governance and management, which accurately capture the dimensions of school performance valued by parents.

3. Middle class exit from the public system is a serious concern. The implications of middle class flight includes a likely further deterioration of quality as pressures on teachers from educated parents are removed; over the longer term, it is possible to anticipate a declining political priority for public education policy and financing.

4. A fuller understanding of the diversity and complexity of the religious education sub-sectors and the phenomenon of the apparently growing preference for religious


vi

education requires further empirical work and analysis. The experience of this study is that such work should be feasible.

5. Evaluating and understanding religious education choices may take into account recent evidence from CAMPE (2008) and others which suggests investments in and returns from madrassah education may be low. However, the choice of religious education may be a form of social progress for poor households who had gained no formal education previously.

6. Public policy needs to gain a greater understanding of the significance of the growth of private education, particularly at primary, with respect to critical issues of equity, financing and quality.


vii

Table of contents

Acknowledgements i

Executive summary ii

Background: the changing demand for education in Bangladesh ii

Global debates on School Choice ii

Methods and approach ii

Findings iii

List of tables and figures ix

Abbreviations xi

1 Introduction 1

1.1 The changing demand for education in Bangladesh 1

1.2 Global debates on School Choice 2

1.3 Report structure 3

2 Methods and study design 5

2.1 The approach 5

2.2 Site selection 5

2.3 School survey 7

2.4 Household research 8

2.5 Community-level research 9

2.6 Analysis and limitations of the approach 13

3 School Supply 14

3.1 School numbers and types 14

3.2 Quality: contact time, teachers, attainment, and facilities 19

3.3 Access and cost 23

3.4 Governance, management and community participation 25

4 Aspirations and attitudes towards education 28

4.1 Education and attitudinal change 28

4.2 Aspirations and expectations 31

4.3 Perceptions of schooling characteristics: quality, access and religiosity 34

5 Schooling choice and outcomes 39

5.1 Actual schooling choices 39

5.2 Schooling choice patterns within households: stylised facts analysis 45

5.3 Outcomes of school decisions 48

6 Discussion and conclusions 53

6.1 School supply 53

6.2 Preferences and aspirations 54

6.3 Actual schooling choices and outcomes 56

6.4 Policy implications 57

References 59


viii

Annex A 61

A.1 Tables 61

A.2 The study sites 66

A.3 Participatory school mapping and timeline exercise 70 A4 Annex note on madrassahs


ix

List of tables and figures

Table 2.1 Summary characteristics of the community case study sites 11

Table 3.1 Distribution of registered schools and enrolment across primary institution types (% of total) 15

Table 3.2 Distribution of registered institutions and enrolment across secondary institutions (% of total) 15

Table 3.3 Numbers of different types of institution, by site 17

Table 3.4 Contact time 20

Table 3.5 Teaching staff by registered/unregistered institution (%) 21

Table 3.6 Repetition and examination performance by area 22

Table 3.7 School physical facilities by registered/unregistered institution (% of institutions) 23

Table 3.8 Accessibility of institutions from the case study community 24

Table 3.9 Average annual household spending by area, poverty level and category (Taka) 25

Table 3.10 Parent Teacher Associations 26

Table 5.1 Educational institutions ever attended by children according to poverty quartiles 40

Table 5.2 Fathers‟ educational level and children‟s primary school types 41

Table 5.3 Mothers‟ educational level and children‟s primary school types 42

Table 5.4 Fathers‟ school types and children‟s primary school types 43

Table 5.5 Mothers‟ school types and children‟s primary school types 43

Table 5.6 Preferences for different forms of government 51

Table A.1 Average number of education institutions in rural upazilas¹ 61

Table A.2 Average fees and other school charges per annum by school type and area (Taka) 62

Table A.3 Average annual household spending by poverty ranking and school type (Taka) 63

Table A.4 Male employment outcomes by education level (%) 64

Table A.5 Female employment outcomes by education level (%) 65

Figure 2.1 Poverty profile of the case study households 9

Figure 3.1 Average number of institutions in rural upazilas and study sites 16

Figure 4.1 What do you hope your children will learn or gain from their education? (% of households, unprompted responses) 30

Figure 4.2 Likes and dislikes about schools ever attended (% of responses) 37

Figure 5.1 Reasons for selecting current school 39

Figure 5.2 Primary school types attended by male children of households 44

Figure 5.3 Primary school types attended by female children of households 45

Figure 5.4 Beliefs about democracy (% responding): features of democracy liked most 51


x

Figure 5.5 Beliefs about democracy (% responding): features of democracy liked least 52

Figure 5.6 Views on the treatment of citizens (% agreeing or strongly agreeing) 52


xi

Abbreviations

AEM Attached ebtedayee madrassah

BRAC (Formerly) Bangladesh Rural Advancement Committee

DPE Directorate of Primary Education

EM Ebtedayee madrassah

EMIS Education Management Information System

GPS Government primary school

IDS Institute of Development Studies

KG Kindergarten school

MOPME Ministry of Primary and Mass Education

NGO Non-governmental organisation

NRNGPS Non-registered non-government primary school

OPM Oxford Policy Management

PTA Parent Teacher Association

RNGPS Registered non-government primary school

RNGSS Recognised non-government secondary school

SMC School Management Committee

SSC-HSC Secondary school certificate/higher school certificate

UNICEF United Nations Children‟s Fund

1

1 Introduction

This report presents findings from the School Choice in Bangladesh study, a qualitative study conducted in four upazilas in Kurigram, Kushtia, Gazipur and Chittagong districts

in 2008. The study aimed to analyse household choices around basic education1 by

exploring a) the diversity and quality of the supply of schools; b) preferences and aspirations with respect to education; and c) schooling decisions and experiences, including the economic, social and political outcomes from education. The research was designed to take explicit account of gender, poverty, economic opportunity, religiosity, and human security as factors affecting decisions around educational investment.

1.1 The changing demand for education in Bangladesh

The motivation behind the School Choice in Bangladesh study was to inform public policy on basic education in a context in which the demand for schooling in Bangladesh has been changing. Enrolment in government and registered non-government schools at primary declined in the first half of the 2000s, while at secondary level, expansion slowed and may now also have begun to decline. These declines were driven by declining enrolments among boys, with some 0.7 million fewer boys enrolled in government and registered non-government schools in 2005 compared to 2000. Enrolment in all types of school stagnated over the period, however, suggesting that many children were transferring to other types of school. Other features of the changing context included the growth of significant gender differences in education enrolment; these are widest among the poorest (Al-Samarrai 2006, 2008).

The evidence indicates that the changing demand for schooling is likely to substantially reflect changes in the types of schools being enrolled in. Yet there is little information available on a) what is driving changing preferences for school types; b) the types of education being provided; and c) the potential outcomes that these institutions can be expected to deliver.

One possible driving force behind changing preferences for school types may relate to concerns that the massive expansion of school provision in the 1990s reduced the quality of education provision. Declining quality in education provision is likely to squeeze two groups out of public schools: the poorest, for whom the low returns from low quality schooling are likely to outweigh the low cost and other incentives to attend public schools; and the rich, who have the capacity to pay for higher quality education. This raises questions about whether Bangladesh is witnessing a situation in which it is increasingly the moderate poor and slightly better off households that benefit from government and registered non-government schools. If so, wider objectives of equity and quality in public education provision, including those set out within PEDP II and the

1 Basic education is defined here as the first ten years of schooling. Within the Bangladesh

education system, this comprises primary (classes 1-5, for ages 6 to 10) and lower secondary (classes 6-10, ages 11-16, ending with the Secondary School Certificate examination). Although the study did not focus on pre-primary or higher secondary (classes 11-12, aged17-18, ending with the Higher Secondary School Certificate examination), higher education institutions which also contained primary or secondary education sections were studied.

2

PRSP, indicate the importance of exploring whether and which dimensions of education quality are indeed important to households in their schooling choice decisions.

Concerns about dropout and non-enrolment among the poorest also persist, as enrolment rates and attainment levels amongst the poorest boys have failed to keep pace with those among boys from better off households (Al-Samarrai 2008a). Anecdotal evidence suggests that poor children may be increasingly attracted to NGO schools, some (but not all) of which are seen as providing a better standard of education. Concerns about rising Islamic militancy in Bangladesh have also directed attention to the growth of madrassah education, through a range of institutions predominantly or partly based on the teaching of Islam and learning in Arabic. Debates about the rise and popularity of madrassah education have to date been informed more by alarmist and journalistic accounts than evidence-based analysis of what is being provided, and by whom.

In this context, the present study was geared towards gathering information on:

a. factors driving changing preferences for schools of different type;

b. changes in the types and quality of education being provided; and

c. expectations of, and as far as possible actual outcomes that these institutions deliver.

The purpose of gathering this information is to enable policymakers to develop an informed response to this changing situation, including by highlighting issues on which further data and analysis are needed.

1.2 Global debates on School Choice

The School Choice in Bangladesh study is also framed by wider global debates about the governance and quality of education service provision. Proponents of school choice argue that enabling parents to select between different types of education providers, typically through voucher schemes, unleash market-like competitive pressures on schools that result in higher quality. In terms of debates about the role of accountability in determining the quality of service provision, school choice is understood to strengthen the link in the accountability chain between service-users (parents) and frontline service-providers (schools/teachers). Debates about school choice have featured predominantly in education policy debates in Sweden, the US and the UK, although drawing on the experience of Chile with public financing for private and semi-private schools since the 1980s, debates about the role of school choice in raising education quality are also increasingly featuring on policy agendas in developing countries.

Debates about school choice have been politically and ideologically heated, yet evidence that choice in education „markets‟ has raised the quality of education in terms

of learning outcomes has to date been scanty and ambiguous.2 In the Bangladesh

context, however, the school choice debate can help to frame analysis of the following:

2 See Levin and Belfield (2004).

3

Diversification and expansion of education service provision. Official data collection has to date focused on the public education system, and little evidence has been available on trends in the diversification and expansion of education service provision in Bangladesh. The study will offer preliminary insights into the direction and character of change in education services.

Accountability and governance in education services. While the evidence that increased school choice has led to improve learner outcomes elsewhere has been weak, there is some evidence that the potential for competitive pressures has helped to raise

standards in some contexts.3 The study of school choice in Bangladesh may help to

isolate more precisely the factors that parents and children believe to affect accountability and governance in education provision at the local level, and thereby contribute to understanding the priorities for education governance reform.

Perceptions of quality in education provision. Changing school preferences and decisions are likely to reflect changing perceptions of the value of different types of school. Exploring school choice issues may contribute to understanding the basis for demand for quality in education services, which is of relevance to wider debates about accountability and governance in education. This may also shed light on the determinants of demand for religious education, which is of interest in the context of concerns about the potential for religious extremism in Bangladesh.

The impact on equity. Evidence from elsewhere indicates that increased school choice leads to „skimming‟ and increased social differentiation in access to education. Exploring the factors shaping school choice in Bangladesh may help to show how and whether poorer children are being left behind in lower quality institutions, while those who can afford to, opt out of the public system. The study can help to inform policies to address and reverse the unequalising consequences of the expansion and diversification of school provision.

1.3 Report structure

The report is structured as follows: Section 2 describes the study design and methodologies, outlining the rationale for the methods adopted, the limitations of the design, and the instruments used. Section 3 presents findings on School Supply in the four study areas, with sub-sections on the provision of schools of different types, comparing the quality of educational institutions surveyed, governance and accountability conditions, and costs of attendance. Section 4 discusses findings about Educational Preferences and Aspirations among households. Section 5 presents findings about Schooling Choices and their Outcomes. The first part discusses patterns of household choices by gender, over time, in different areas and by socioeconomic status. The second part discusses the evidence on outcomes from different types of education, exploring experiences of households, the impact of community-level factors such as the extent and nature of economic opportunity; and socio-political outcomes, with specific reference to attitudes towards democracy, gender equality and the role of religion in politics and the state. Section 6 concludes with a discussion of the findings and their policy implications. An annex contains additional detail of the methodologies used and data collected.

3 Chakrabarti (2007); Hoxby (2003).

5

2 Methods and study design

2.1 The approach

The approach selected for the study was a comparative case study methodology, involving four contrasting communities. An initial plan to adopt a large-scale or nationally representative survey was rejected, primarily on grounds that this would not necessarily yield an in-depth understanding of the complex processes of household decision-making around education. Instead, a multi-disciplinary, multi-method strategy for approaching these issues capable of exploring the issues at individual, household, community, school and regional levels was considered more appropriate to the questions being addressed in the research. The approach adopted was to select four villages in four distinctly different districts, and to use the same research instruments to study individual, household, community, regional and school characteristics in each.

Two research teams each composed of a school expert, a lead community researcher, and four additional researchers spent between one and two weeks researching each site. Each team had two women researchers, to enable entry to girls‟ schools and to interview women in the more conservative households. Each team was visited by the lead researcher.

Instruments were pre-tested in Kishoreganj, revised and piloted in Gazipur, and the researchers received a week‟s worth of training. Quality control and supervision was provided by the school expert and lead community researcher in each team, with support from the team leader.

2.2 Site selection

Based on knowledge of the factors shaping household decisions in education in Bangladesh from earlier studies and other sources, four sets of community characteristics were identified as important contextual determinants of school choice:

High levels of poverty and extreme poverty, remote location. This can be expected to shape a) community norms placing a lower than average value on children‟s education; b) lower than average capacity to invest in children‟s education; c) higher than average household need for children‟s labour and d) limited expectations of economic returns on children‟s education.

High economic growth, good employment and livelihood options. These features should entail a) community norms value forms of education that are associated with employment success; b) higher than average capacity to invest in children‟s education; and c) higher than average expectations of economic returns in children‟s education; but in addition, d) greater potential demand for children‟s labour, which is likely to influence educational decision-making.

Relatively insecure in terms of livelihoods and physical wellbeing. Areas where people‟s livelihoods and physical wellbeing are insecure are also areas in which we may expect a) community norms to place a lower than average value on long-term investments in children‟s education; b) any expectations of returns on children‟s

6

education may relate to migration or immigration rather than opportunities within the local economy; and c) strong claims on children‟s labour and early marriage.

Reputation for religiosity, many migrants to Middle East. In such an area, we should expect a) above average demand for religious as compared to secular schooling, both for boys and for girls; and b) expectations that investments in religious schooling will result in social as well as economic returns.

Communities were selected through a three-stage process.

1. National-level data were drawn on to select the four districts. Kurigram was selected for the poor rural community, based on poverty statistics and the research team‟s previous research on education in that district. Gazipur was selected as the area of high economic opportunity, based on its proximity to Dhaka and the presence of a cluster of readymade garments factories in that district. National data on crime and violence were used to select Kushtia district as the insecure region. Chittagong was selected on the basis of earlier research which had identified it as a centre for Islamic education, and on the basis of knowledge of its greater religiosity and accompanying social attitudes.

2. Analysis of education provision based on Directorate of Primary Education (DPE) data was undertaken at the upazila level for each district. The purpose was to ensure that each selected upazila was broadly in the mid-range of non-shadar upazilas in terms of provision of different types and numbers of educational institutions, and was therefore roughly representative of the district. The aim was to exclude upazilas that were unusually under-supplied by schools, as this would prevent detailed exploration of school choice issues, as well as to exclude upazilas with a particularly large supply of institutions, such as is often the case in the main district towns or shadar upazilas. Annex 1 contains a graph showing school provision in the selected upazilas in comparison to non-shadar upazilas.

3. Visits to the selected upazilas were undertaken, and brief interviews with upazila education officials and other informants were undertaken to identify specific communities or villages which featured the community characteristics sought, but which were also supplied by a sufficiently wide range of different school types. Visits to the communities themselves were then undertaken to confirm that the communities were suitable representatives of the types sought.

The careful selection process was deemed necessary on the basis of experience that the advantages of comparative analysis could only be gained if the cases were indeed representative of the characteristics under study. In the present study, the selection was successful, and the communities selected proved to feature the anticipated characteristics. One partial exception was that environmental insecurity was not a significant feature of the Kushtia case study, although as the Kurigram site was a chor region, these features were picked up in that case study.

7

2.3 School survey

In each site, all schools providing basic education deemed to be within the catchment area of the community were surveyed. Pre-primary institutions were excluded, although some data on higher secondary institutions was collected where these were part of institutions providing basic education. The definition of catchment area developed was within 2km of the outer boundary of the community for primary, and 10km for secondary. This definition was arrived at on the basis of discussion with education experts and data, and aimed to ensure that all institutions that could be potentially attended through daily travel were included. This excluded institutions at which local children boarded or lodged, although some information about which children do so and in which types of institution was also collected through the qualitative elements of the research. The research found that this definition of the catchment area was generous, and that the overwhelming majority of local children attended schools very close to their homes, particularly at primary. However, the definition permits a comprehensive assessment of the quality and range of educational institutions that could conceivably be seen to be supplied to the communities, and so served the purposes of the research.

The identification of a comprehensive list of educational institutions4 serving each

community involved the following steps:

1. visits to the upazila education office, from which lists of registered and recognised schools were gathered5;

2. visits to community leaders and schools in the selected communities, to identify other unrecognised institution; and

3. a participatory school mapping exercise, in which children and educated youth mapped out the institutions attended by local children.

Between 15 and 44 institutions were identified in each of the four sites, resulting in a total of 106 institutions.

Each school was surveyed using a questionnaire designed to collect data on the following issues:

Background characteristics: year of establishment, official status;

Facilities;

Teaching and learning;

Governance and management; and

Student performance and outcomes.

4 This report refers to „educational institutions‟ and „schools‟ interchangeably, although the term

„school‟ is not usually used to refer to madrassahs in Bangladesh.

5 From here-in registered is used for simplicity to describe either registered primary schools or

recognised secondary schools.

8

2.4 Household research

In each community, a total of 30 households were interviewed in depth. Six were selected from among households whose children had dropped out, never enrolled, or which had a lower overall educational attainment than the rest of the community. The remaining 24 were selected from students attending the schools serving the area. The research design aimed to sample children as evenly as possible from each of the schools in the area, although as noted above, in many of the more distant institutions, none of the community children were enrolled.

Once children had been sampled, the households were visited, and interviews undertaken with the household head or spouse, and short interviews were also undertaken with up to two children in the household.

The household interviews contained a significant number of open-ended questions, and were designed to collect information on:

Demographic and household composition;

Poverty status;

Income and occupation;

Education history and experiences;

Factors affecting school choice such as perceived quality and preferences for different types of institution;

Cost of education;

Current enrolment and attendance; and

Aspirations and expected outcomes.

Figure 2.1 summarises the poverty status of the case study households. Using a poverty score card developed on the basis of the HIES 2001, household responses to questions about their poverty status were scored, and ranked into quartiles on the basis of their likelihood of being poor. An important methodological issue to bear in mind throughout the report is that the households were sampled on the basis of children attending schools within the school sample defined by the catchment area approach, identified through the school listing exercise. This means that the households are not necessarily representative of their communities. In addition, in each community six households were selected to represent households with a higher propensity to dropout or not enrol their children in school. The poverty profile of the interviewed households is, nevertheless, generally consistent with what is known of the broader socioeconomic conditions of the communities. The highest number of poor and the lowest number of rich households were found in Kurigram: this makes sense because the Kurigram site was selected as the example of a „poor‟ community. After Kurigram, Kushtia had the next fewest rich households, but of all the four communities, it also had the smallest number of households in the poorest quartile, with only four.

9

Figure 2.1 Poverty profile of the case study households

Notes: 1) source: household survey, School Choice study.

After Kurigram, the community with the most households in the bottom half was the Chittagong community, from which 9 households in the richest quartile were also selected. Overall, the Chittagong community and its surrounding area were distinctly wealthier than the Kushtia site. However, two factors about the Chittagong community help to explain why our sampled households seem poorer than in Kushtia. First, many households in the Chittagong community, including some in our sample, had recently experienced sharp income downturns with the closure of the illegal timber trade under the caretaker government regime. Second, the strategy for sampling households meant sampling as equally as possible from all educational institutions within the school sample: given the very large number of semi-formal madrassahs in the Chittagong area, which poorer boys are expected to attend, it is likely that our household sample is slightly more biased towards poorer households there than in the other sites, where fewer madrassahs were found. The most households in the upper half of the poverty distribution were in Gazipur, which was fully consistent with the socioeconomic conditions of that community. Again, however, our school-based sampling strategy may have meant a slight bias towards poorer households, as the poor seem more likely to attend schools that are close by, and the qualitative evidence revealed that some richer households in Gazipur send their children to prestigious schools in Dhaka, from which we did not sample.

2.5 Community-level research

To deepen understanding of the regional and community-level factors such as economic opportunity, attitudes towards gender, education, religion and children‟s work, community-level research was also undertaken. This involved the following:

10

Introductory meetings with community leaders, to build acceptance of the research team, gather views on village history, economic conditions, and attitudes towards investment in children.

A participatory school mapping exercise in each area, to enable the participation of children and young people in the research. This exercise was oriented towards producing maps of all types of educational institutions in the area, a timeline to document the historical establishment of schools in the community, and discussion of preferences and constraints in the choice of schools.

Three focus group discussions in each area, with members of a local SMC, women and men. These discussions covered the supply of schools, perceptions of quality, preferences for different types of institutions, aspirations and expected outcomes, and change over time in education provision and demand.

Two additional in-depth case studies in each area, involving follow-up interviews with households identified as having particularly interesting or illustrative experiences of school choice and educational investment practices.

English translations of the instruments used, and details of the study sites are provided in the annex. Table 2.1 provides summary information about the community case study sites.

11

Table 2.1 Summary characteristics of the community case study sites

Site Poverty and social structure

Economic and employment opportunity

School provision and attitudes to education

Other factors

Bhurungamari, Kurigram

- Land ownership concentrated among 10 per cent richest - high proportion of extreme poor HHs - poorer HHs recently settled - high proportion of landless - 80 per cent depend on wage labour - Women from ultra poor HHs involved in wage work - child labour routine among ultra poor - early girl and boy marriage common

- Most rely on agricultural wage labour, c. 100 days per year - Much land is sandy chor, difficult to cultivate - Irrigation from 1990s, padi production grew - Jute production significant - Rickshaw pulling and small business also common - Seasonal migration among poor - Semi-permanent in-country migration to Comilla, Dhaka - recently started to send children to work in garments

- One GPS, difficult to maintain because of river erosion, reluctance to donate land - Environmental insecurity encourages reliance on child labour - Rich limit educational investment to secondary school, withdraw boys to manage land - Popular madrassah forced to close because Imam’s Tk 400 salary could not be raised

- chor area, subject to annual flooding, riverbank erosion - Difficult to establish land title - Excluded from Union Parishad political settlements - Little social mobilisation - Few NGOs - Roads improved recently

Kumarkhali, Kushtia

- land ownership concentrated among 3 richest HHs - rich Hindu HHs mostly in India, some land and relatives remain

- predominantly dependent on agriculture, small farms - sharecropping/mortgaging by landowning HHs common - many have small businesses - minority have formal employment

- landlord families uninterested in education - GPS and a madrassah in village since 1930. 10 new institutions since 1990 - madrassah closed, teachers returned to home districts

- highly insecure region until RAB/Caretaker Government - social cohesion despite religious difference - rich cultural heritage and tradition, with mazars and links to Tagore

Kaligonj, Gazipur

- wealthy area, few poor: small proportion dependent landless - poorest and upper middle groups women work - very little child labour - 10 per cent small landowners - 20 per cent rich landed

- poor minority rely on wage labour, incl. women, rickshaw, agric labour - 10 per cent small landowners, women do not work, some formal jobs - 20 per cent rich landowning

- one para where education levels are low, children work - overall strong preference for education - after independence no schools, had to travel far for education - rights awareness after

- NGOs since 1980s - little social organisation or local civil society

12

- majority richest combine landowning, business and migration, women and children do not work

HHs professionals e.g. lawyers, civil servants, incl. women teachers - majority richest group manage land, have businesses - insufficient labour for agricultural production, seasonal in-migrants from poor north - middle class sons migrate as soon as old enough

independence meant rising demand for education - schools increased in 1990s - most now educate up to jr secondary - since 2000, poorest enrolled children in new NGO schools - girls’ marriageable age risen since 1990, with girls’ secondary scholarship, to ‘after SSC’

Rangunia, Chittagong

- generally wealthy area, few poor - 60 per cent Muslim, 40 per cent Hindu and adivasi communities - many claim foreign ancestry - Muslims divided into Sunni and Wahabi, previously conflictual

- hilly region, vegetable and fruit production high - lowland mainly agricultural, highly productive padi - c. 20 per cent migrants to Middle East, help others to migrate - small and medium business common, particularly timber - women run small businesses with NGO loans - adivasi women grow and market produce

- Muslim children first study in maktab or madrassah - strong preference for madrassah education - Muslim and advisasi children study in same schools, mix freely - local madrassahs financed by donations from migrants in Middle East

- many mazars, mosques and madrassahs - protected against floods, irrigated land since 1970s’ canal-building programme - illegal timber trade hit hard by Caretaker Govt crackdown - many NGOs

Notes: 1) source: interviews with key informants, School Choice study.

13

2.6 Analysis and limitations of the approach

Data collected from the school and household surveys were entered, cleaned and processed, using the Stata data analysis package. The qualitative research findings were processed in Bangla, and the qualitative and quantitative results were analysed in conjunction, in order to triangulate and enrich understanding of the findings.

The main limitation of the approach is that because the findings are not based on a nationally representative household survey, they are not representative of factors driving school choice across Bangladesh as a whole. In addition, the households were purposively sampled and not randomly selected, and therefore are not representative of their communities.

However, the approach permitted a depth and comprehensiveness to the understanding of the issue that large-scale survey methods alone would not yield. Used with caution, the analysis of the findings permit a degree of generalisation about the factors and constraints shaping school choice in contexts of extreme poverty, insecurity, diverse formal sector economic opportunity, international migration, and in relation to religious belief and practice, within Bangladesh.

14

3 School Supply

This section presents findings on the provision of education in the study areas. It starts by situating the local supply of schools within the national context. It goes on to explore findings about the quality of the institutions available, with specific reference to teaching and learning inputs and outcomes and school facilities. This is followed by findings about the cost and distance of institutions from the study areas. The section concludes with findings about the governance, management and community participation in the institutions.

3.1 School numbers and types

3.1.1 National estimates

National statistics on the full range of school types are not available in Bangladesh. DPE‟s EMIS collects data on 10 types of registered primary-level institutions which follow the government‟s curriculum and structure, and which the Ministry of Primary and Mass Education (MOPME) recognises:

government primary schools (GPS);

registered non-government primary schools (RNGPS);

non-registered non-government primary schools (NRNGPS);

experimental schools;

ebtedayee (primary equivalent) madrassahs (EM);

kindergarten 1;

kindergarten 2 (NGO schools);

community schools;

attached ebtedayee (primary equivalent madrassahs (AEM);

(primary sections of) attached high schools.

Table 3.1 shows the distribution of registered institutions and enrolment across primary institutions by type. More than 72 per cent of all registered primary institutions are either government primary or registered non-government primary schools, enrolling more than 81 per cent of all children in registered institutions. Of the remaining groups, only the ebtedayee and attached ebtedayee madrassahs enrol notable proportions of children, with 11.7 per cent of children in nearly 18 per cent of all registered primary institutions.

Table 3.2 shows the distribution of registered institutions and enrolment across secondary institutions by type, excluding basic education provided in institutions that provide education above higher secondary level, such as colleges. The secondary sub-sector is dominated by the registered non-government secondary schools (RNGSS) providing junior and full secondary (up to the Higher Secondary Certificate level, class 12, age 18), which constitute 69 per cent of all the institutions at this level, and 70 per cent of all enrolment. Dakhil and alim madrassahs within the government-supported Aliya madrassah system constitute nearly one-third of all institutions at this level, and over a quarter of enrolment, at 27 per cent. Government secondary schools are a minuscule 1 per cent of all institutions, enrolling only an elite 2 per cent of all students at this level.

15

Types of institution that are excluded from official primary statistics include the majority of NGO schools, such as the 32,000 BRAC primary schools which were enrolling 984,000

students in 2007.6 Other NGO schools are also excluded from official statistics, including

institutions supported under BRAC‟s Education Support Programme, as well as the Dhaka Ahsania Mission and FIVDB schools.

Table 3.1 Distribution of registered schools and enrolment across primary institution types (% of total)

Notes: 1) source: Directorate of Primary Education, PEDP II baseline survey 2005.

Table 3.2 Distribution of registered institutions and enrolment across secondary institutions (% of total)

Institution type Institutions (% of total) Enrolment (% of total)

Govt Secondary 1.2 2.2

Non-govt Junior Secondary 16.3 8.9

Non-govt Secondary 52.3 61.5

Dakhil Madrassah ¹ 25.2 22.0

Alim Madrassah ¹ 5.0 5.4

Notes: 1) Refers to officially registered and supported non-government madrassahs in the Aliya system. Dakhil madrassahs are equivalent to Junior Secondary, and Alim to full Secondary. 2) source: BANBEIS data for 2005.

A second set of important and large-scale providers excluded from official figures is the quomi madrassahs, private madrassahs following a different system and curriculum to the government-financed aliya madrassah system. No official estimates of the numbers of quomi madrassahs or enrolled students exist, although the quomi umbrella organisation reportedly claims some 15,000 such institutions are registered with them, and a further estimated 64,000 unregistered madrassahs of various types are also estimated to be operating (Bano 2007). However, these 64,000 madrassahs are likely to include the many thousands of small maktabs (Quran reading classes attached to mosques) and hafeziya madrassahs, which teach Arabic, but which do not resemble the more formalised education systems of either the Aliya or the quomi madrassah streams.

6 BRAC Annual Report 2007, Dhaka: BRAC.

Institution type Institutions (% of total) Enrolment (% of total)

GPS 47.8 59.6

RNGPS 24.3 21.8

NRNGPS 1.2 1.0

Experimental 0.1 0.1

Ebtedayee madrassah 7.8 4.8

Kindergarten 1 2.8 1.5

Kindergarten 2 (NGO) 0.3 0.2

Community 4.0 2.8

Attached ebtedayee madrassah

10.1 6.9

Attached high school 1.5 1.5

16

Other institutions excluded from official records are private schools, including English-medium schools, kindergarten schools which do not follow the national curriculum and structure, and coaching centres, distinct from private tuition in that teaching takes places in an institutionalised format, often very similar to school-based teaching. Little data is available on the extent or distribution of any of these private institutions, although anecdotal evidence and cost considerations suggest these are likely to be concentrated in the larger cities and wealthier towns.

3.1.2 Upazila-level provision

To the extent possible, and based on the data available, the study areas were selected to represent average levels of provision of different school types and numbers. Efforts were made to ensure that at least five types of institution were available at primary level for each area. While this meant excluding some particularly remote areas in which the numbers and types of schools were more limited, this was considered necessary to enable a realistic exploration of factors affecting school choice for the majority.

Figure 3.1 presents estimates of the average numbers of different types of educational institutions in rural upazilas for which official statistics exist. It suggests that for the dominant types of institutions, the selected upazilas are neither unusually under-supplied, nor excessively over-supplied. Estimates in Figure 3.1 should be treated as illustrative and not as definitive data sources, as the official data contain a number of gaps and inconsistencies.

Figure 3.1 Average number of institutions in rural upazilas and study sites

0

10

20

30

40

50

60

70

80

90

100

GP

S

RN

GP

S

NR

NG

PS

4.E

M

Kin

de

rga

rte

n 1

Kin

de

rga

rte

n 2

(NG

O)

Co

mm

un

ity

AE

M

Att

ach

ed

Hig

h S

cho

ol

Se

con

da

ry

Ma

dra

ssa

hs

National

Bhurungumari,

Kurigram

Kumarkhali,

Kushtia

Kaligonj,

Gazipur

Rangunia,

Chittagong

Notes: 1) Metropolitan and shadar upazilas are excluded. 2) Source: primary institutions: Directorate of Primary

Education, PEDP II baseline survey 2005 and secondary institutions: BANBEIS data for 2005.

17

3.1.3 Local provision

An extensive, multi-method approach was taken to identifying the full range of educational institutions serving the study populations. Table 3.3 provides summary information of the numbers of different institutions in each of the four areas, based on the school listing exercise discussed in Section 2.

Table 3.3 Numbers of different types of institution, by site

Educational Institutions Kurigram Gazipur Kushtia Chittagong All

Primary Institutions

Government primary school 1 5 3 5 14

Registered non-government primary school

1 2 1 1 5

NGO primary school 6 3 1 1 11

Community primary school 0 1 0 0 1

Kindergarten school 0 2 0 1 3

Ebtedayee section of dakhil, alim or fazil madrassah within catchment area (less than 2km) ¹

0 0 1 1 2

Primary-level non-aliya madrassah 0 1 0 2 3

Hybrid primary school 0 0 0 1 1

Secondary Institutions 0

Government secondary school 0 0 0 1 1

Registered non-government secondary school

4 8 5 14 31

Registered non-government junior secondary school

1 1 2 2 6

Coaching centre 0 0 0 1 1

Dakhil madrassah 5 1 2 2 10

Quomi secondary madrassah 0 0 0 1 1

Higher Institutions (but offer secondary and/or primary classes)

0

Registered non-government intermediate college

0 0 0 1 1

Alim madrassah (including primary sections)

0 2 1 2 5

Fazil madrassah (including primary sections)

1 1 0 1 3

Other 0

Other madrassah (unregistered by aliya or quomi authorities; semi-formal)

0 1 0 8 9

Total 19 28 15 44 106

Note: 1) Ebtedayee sections of dakhil, alim or fazil madrassahs are not included in the total as they are accounted for in the secondary/higher institutions. 2) Includes three schools that were not available for interview. 3) „Hybrid‟ category includes an institution which features Islamic education but also shares features of Kindergarten schools. 4) Source: school listing exercise, School Choice study.

18

Table 3.3 provides important findings about the diversity of schools being supplied to the different communities. At primary, the poorer communities of Kurigram and Kushtia can in effect choose from four types of institution, or three, if government primary and registered non-government primary schools are treated as the same type of institution, which in key respects they are. This includes the ebtedayee sections of local aliya madrassahs. Gazipur and Chittagong both boast no fewer than six types of primary institution within 2km of the community, as well as each having at least one semi-formal primary-level madrassah available, providing religious education in Arabic in the evenings. One primary institution, in Chittagong, was a formal school with an emphasis on religious education, which also taught in Bangla. This institution has been categorised for our purposes as „hybrid‟. Other new institutions exhibited „hybrid‟ characteristics, particularly a strong emphasis on religion alongside privately taught general education, but other characteristics of these made it possible to classify them as either KG schools or madrassahs.

At secondary, the Kurigram and Kushtia communities face choices of registered non-government junior or full secondary schools, or of aliya madrassahs going to Dakhil or Fazil (Kurigram) and Dakhil or Alim (Kushtia) levels. This entails a choice set of two types of institutions in terms of curriculum and officially-registered type. Kurigram had 11 and Kushtia 10 secondary institutions within the defined catchment area. In Gazipur the diversity of secondary school types was in effect not much greater, although that community could choose between aliya madrassahs going to Dakhil, Alim and Fazil levels. But in terms of actual numbers, there were 13 secondary institutions within the area, and the qualitative research revealed that many children from wealthier families also commute into Dhaka to attend more prestigious institutions, suggesting that the actual choice set for wealthier households is significantly wider than the institutions within the defined catchment area. The Chittagong area had the most diverse and the most numerous secondary institutions, boasting one of the country‟s just over 200 prestigious, well-resourced government secondary schools, as well as registered non-government secondary and junior secondary schools, a registered non-government college with a secondary section, aliya madrassahs up to Dakhil, Alim and Fazil levels, and the only formal quomi madrassah in the entire sample. This meant a total of 24 secondary-level institutions.

There were also an additional nine semi-formal madrassah institutions in Chittagong, and one in Gazipur, none of which were part of the registered aliya system, nor did they appear to be part of the formal but unregistered quomi madrassah system. While these institutions exclusively teach religious education and Arabic language, and the medium of instruction is Arabic, most had semi-formal institutional characteristics. The school listing exercise excluded the informal maktabs attached to mosques that commonly teach religious studies to children before they attend schools, and so these 10 semi-formal but unregistered madrassahs were not of this type. All featured „madrassah‟ in their names, and most appeared to be substitutes for or alternatives to general education. As there was little prior information available about these institutions, the study instruments were not necessarily well-tailored to capturing all aspects of the services they provide. However, the data that was collected indicates that they provide a fairly extensive, semi-formal system of religious education. Only two children within the communities could be identified as attending these

institutions;7 neither were attending other institutions in addition to their evening attendance

at the madrassah, and both reported ambitions to study up to Masters-equivalent level.

7 This reflects the fact that the catchment area was defined to be sufficiently broad to capture the full

potential choice set for each community. This meant that not all institutions surveyed in fact had students from the communities they in theory serve.

19

Of the NGO schools, 9 out of 11 were BRAC schools, which are the single cohort non-formal schools which educate between 30 and 33 students, around 60 per cent girls from poor families.

Although there were only three kindergarten (KG) schools within the total sample, this type of institution had earned a positive reputation for education provision of high quality which extended beyond the number of households with any actual experience of them.

An important finding about school supply is that many institutions, in particular NGO and KG schools and many madrassahs, are comparatively new. The popular perception, from discussions within case study communities, is that the number and diversity of schools has increased considerably since the 1990s. See Annex 3 for the results of participatory timelines intended to explore how school supply had changed over the last generation,

undertaken with children, youth and other community members.8

3.2 Quality: contact time, teachers, attainment, and facilities

Debates about school choice frequently focus on the issue of the quality of instruction, with a central claim of proponents of wider choice being that this creates competitive pressures which ultimately produce higher quality education. The present study does not provide strong quantitative support for this hypothesis, for two reasons. First, because there is little quantifiable evidence based on commonly used indicators of quality that newer entrants into education service provision are of higher quality. Second, there is little quantifiable evidence that communities with more diverse and more numerous schools are better served in terms of the quality of education that they receive than communities with few schools of a limited range of types. There are some qualitative findings which support an argument that increasing school choice leads to a degree of competitive pressure which improves some elements of performance; conversely, there are also qualitative findings that suggest wider choice leads to increasing differentiation between socioeconomic groups. Overall, an important issue to bear in mind is that the standard indicators of quality do not necessarily closely reflect the features of schools that people prefer and take into account when selecting an institution.

Table 3.4 presents findings, using proxy indicators, on the amount of contact between students and teachers. In terms of pupil-teacher and pupil-class ratios (or section sizes), the evidence indicates that the newer types of institutions offer more individual contact. Teachers teach on average 50 children in each GPS class section in the present sample, and around 34 in each RNGPS; both of these figures are similar to the findings of the SSPS surveys (FMRP primary 2006; secondary 2006) - which were nationally representative sample surveys. The equivalent figures are lower for NGO and kindergarten schools, as well as the one hybrid-type of primary school, with primary-level madrassahs ranking in between. At secondary, too, the madrassahs have somewhat more favourable pupil-teacher and pupil-

8 It should be noted that the institutions identified through this process were not identical with those

identified as serving the community under the definition of catchment area and the school listing exercises used in the present study. The children, youth and other community members who participated in the school mapping and timeline exercises mainly identified institutions that fell within the geographical boundaries of the village – not all of which necessarily fell within the catchment area definition, or which local children were generally known to attend. This included some categories of institutions which the present study did not include, such as private tutors, pre-primary and post-basic education institutions.

20

class ratios or section sizes than any of the registered general stream secondary-level institutions. The non-formal madrassahs are at the higher end of the pupil-teacher ratio, with 43 students per teacher, although in terms of section size, they are more favourable than the registered general stream primary or secondary schools.

Table 3.4 Contact time

Educational Institutions Average number of days institution was open Jan-Jun 2008

Average pupil-class ratio (section size)

Primary

Government primary school 122 50 (47)


124 34 (40)

NGO primary school 118 27

Community primary school 129 26

Kindergarten 129 15

Primary madrassah 132 36 (33)

Hybrid primary school 103 15

Secondary

Government secondary school 79 48


116 58


100 47

Dakhil madrassah 121 28

Secondary Quomi madrassah 121 32

Higher

Registered non-govt intermediate college

115 39

Alim madrassah 123 37

Fazil madrassah 110 38

Non-formal madrassah

non-formal madrassah 136 43

Notes: 1) Figures in brackets represent comparable findings from the SSPS surveys (FMRP 2005; 2006). 2) source: institution survey, School Choice study.

In terms of the numbers of days the different types of institutions are open, the picture is more mixed: the one hybrid primary school was open only 103 days in the first six months of the year, compared to between 122 and 132 for other types of institution, and 118 for the NGO schools. The range was far wider at secondary, with the general stream schools on average open far fewer days than any of the madrassah institutions. The two government

21

secondary schools were only open on average for 79 days, compared to a high of 136 days for the non-formal madrassahs.

Table 3.5 Teaching staff by registered/unregistered institution (%)

Teacher characteristics Registered institutions

Unregistered institutions

Teachers educated above HSC level

96 81

Teachers with professional teaching qualification

51 2

Teachers with at least 10 years teaching experience

62 25

Women teachers as a percentage of all teachers

25 25

Teachers absent on day of survey

12 (Secondary: 10; GPS 16; RNGPS 11) *

5

Notes: 1) Figures in brackets represent comparable findings from the SSPS surveys (FMRP 2005; FMRP 2006). 2) source: institution survey, School Choice study.

In terms of the quality of teaching and teaching inputs, Table 3.5 suggests that teachers in schools that are formally registered were significantly better endowed with formal qualifications and experience than in other schools in the total sample. This suggests that newer types of institutions, fewer of which are registered, are less well-endowed in terms of teacher qualifications, than more established school types.

By contrast, teachers in the unregistered types of institution were more likely to be present. While some 12 per cent of teachers in registered institutions were absent (for reasons that included legitimate or official absences) on the day of the survey team visit, the equivalent figure was only 5 per cent for the other institutions.

The claim that wider choice impacts positively on school quality is also contradicted by findings about educational attainments in the different areas. Note that there is considerably wider school choice in the Gazipur and Chittagong areas than in Kurigram or Kushtia, whether defined by our catchment area definition, or in terms of practical access (see below). However, as Table 3.6 suggests, the pattern that emerges suggests that attainment levels are highly mixed. Note that out of ten indicators, Kurigram performed best on four, Kushtia on four, Chittagong on two, and Gazipur, with the most diverse range and the largest number of schools on offer, on none. One possibility is that the widening of choice reflected in more and more costly school options may have meant a worsening of inequality, as a process of „cream-skimming‟ occurs, with children from wealthier and better educated households withdrawing from publicly-funded schools to attend more expensive private schools outside the catchment area. There may be some reasons to believe this to have been the case in Gazipur and Chittagong: in some of the less guarded interactions with wealthier community members it emerged that richer households were increasingly reluctant to attend public schools in which poor children were numerous. From one household came the view that it would be a source of great shame and embarrassment should one of their children be selected for the primary education stipend (and therefore signal to the community that they

22

were poor). And the wealthier households in Gazipur were known to send their children to

elite Dhaka schools, lodging with relatives or travelling daily.9

Conventional educational attainment indicators are generally considered difficult to interpret, however; rather than read these findings as strong evidence that wider school choice actually leads to declining educational attainment, it may be worth noting that a) these indicators are based on small sample sizes and that b) they may not, in any case, be strong indicators of the overall quality of the education being provided in a particular institution.

Table 3.6 Repetition and examination performance by area

Average per school: Kurigram Gazipur Kushtia Chittagong

% of students repeating in 2008 7.0 9.2 4.2 2.5

% of students sitting for class 5 scholarship exam in 2007

1.3 1.2 1.4 1.1

No. of students who sat for Class 5 scholarship exam in 2007

13.0 7.3 7.0 6.0

% of exam candidates receiving Class 5 scholarship in 2007

12.9 10.0 19.6 18.9

% of students sitting for class 8 scholarship exam in 2007

2.9 2.4 3.3 2.9

No. of students who sat for Class 8 scholarship exam in 2007

17.9 15.6 16.7 11.8

% of exam candidates receiving Class 8 scholarship in 2007

4.6 1.3 3.5 4.3

% of students sitting for SSC/DM exams in 2007

8.8 7.7 9.0 10.5

No. of students who sat for SSC/DM exams in 2007

53.3 49.5 45.0 43.1

% of candidates passing SSC/DM exams in 2007

57.9 51.3 62.1 52.3

Note: 1) Low figures for the indicators on the % of candidates for scholarship exams reflect that this indicator was based on total enrolment, and not the enrolment in that class. 2) This table refers to students in eligible primary (Class 5 exams) or secondary (class 8, SSC/DM exams) institutions. 3) The per cent of students sitting for each category of exam is an estimate using 2007 figures for exam candidates and enrolment figures from April 2008. 3) Source: institution survey, School Choice study.

Table 3.7 shows that registered institutions outstrip unregistered institutions in terms of school facilities by a considerable extent. As will be discussed below, in Section 4, physical facilities can be important negative features of schools, but they are rarely important elements of what is liked about schools, perhaps particularly among parents. The survey findings are supported by the more in-depth discussions with parents, which suggested that they were likely to weigh up factors such as the quality of school facilities against what they understood about the quality of instruction and care by teachers; persistently strong preferences were expressed for BRAC schools, for instance, despite the widespread dislike of the facilities, with children sitting on mats on the ground, usually in cane and bamboo-

9 Evidence suggests that interventions to widen school choice have at times led to increased

inequality in this manner; UNESCO (2008) Global Monitoring Report, Paris: UNESCO.

23

walled single roomed buildings.10 Similarly, the popularity of KG or kindergarten schools held despite the fact that these are typically housed in buildings that are not purpose-built, which feature small and noisy classrooms, and where there is little outdoor space or playground facilities.

Table 3.7 School physical facilities by registered/unregistered institution (%

of institutions)

Institutions with the following features: Registered institutions

Unregistered institutions

Boundary wall 29 14

Playground 87 18

Brick/cement outer walls 89 64

No noise from adjacent classrooms 45 18

Most classrooms in good state of repair 52 54

Functional toilets for use by students 99 61

Functional toilets for females use only 73 29

Safe water 89 57

Electricity 80 57

Notes: 1) source: institution survey, School Choice study.

3.3 Access and cost

Access and cost appear to be the deciding factors around school choice, presenting serious constraints on the capacities of poorer households, in particular those in more remote areas, to exercise effective choice between institutions.

The ease of access to school does not depend entirely on distance: the condition of the roads in wet and dry seasons, the supply and cost of transport, and the perception of the safety of the route all matter to how people view the issue of access. As Table 3.8 shows, fewer institutions are practically accessible to the Kurigram and Kushtia communities than for the Gazipur and Chittagong sites. Using the distance of half a kilometre to denote a genuinely local or neighbourhood school, Kurigram has one and Kushtia none, compared to Chittagong‟s three, and Gazipur‟s impressive six. In the wet season, primary school-going children can effectively choose between two institutions in Kurigram and four in Kushtia that will take them less than half a hour to travel to, again in distinct contrast to the 12 in Gazipur and 19 in Chittagong. With respect to secondary institutions, only Chittagong had more than one which was within a range of one kilometre. If secondary schoolchildren were able to travel up to an hour in the wet season, however, they would have a reasonably wide choice of institutions, ranging from 8 in Kushtia to 17 in Chittagong.

10

Preferences for BRAC schools were always discussed with reference to the quality of instruction, and there was no suggestion during this research that BRAC school selection was linked to access to microcredit. As BRAC school places are more limited in supply than NGO credit, it seems unlikely that the need for credit drives preferences for BRAC schools; however, this is not impossible, depending on the context.

24

Table 3.8 Accessibility of institutions from the case study community

Number of Institutions

Distance from the centre of the case study community

Kurigram Gazipur Kushtia Chittagong

Primary less than 0.5 km away 1 6 0 3

less than 30 minutes in wet season away

2 12 4 19

Secondary institutions

less than 1km away 0 1 1 5

less than 60 minutes in wet season away

9

10 8 17

Note: 1) By students‟ usual mode of travel 2) source: institution survey, School Choice study.

In terms of the cost of schooling, while average annual household spending on education matched our poverty index-based ranking of the four areas (see Table 3.9), suggesting households are likely to spend as much as they can afford on schooling, when looked at more closely, it seems that some of the poorer households were spending more than some of the richer households on education (Table A.3). The range of household spending was wide, with the lowest average in Kurigram (Tk 3,703) and the highest in Chittagong (Tk 15,462). It should be noted that the high average spending in Chittagong was substantially driven by the high cost of attending the registered private junior college in that area. Spending on madrassah education of all kinds and levels appeared to be high, and with a few exceptions, higher than on general stream education.11

11

For further tables on the cost of education institutions and household expenditure on schooling refer to Table A.2 and Table A.3.

25

Table 3.9 Average annual household spending by area, poverty level and category (Taka)

Kurigram Gazipur Kushtia Chittagong

1st Quartile 1334 887 1833 2805

2nd Quartile 2133 1938 1775 3974

3rd Quartile 2749 2985 2651 5865

4th Quartile 2510 3346 8838 12265

Fees and other charges 250 731 379 2648

Transport to and from the institution

289 319 673 1365

Learning materials (textbooks, stationery, etc.)

1120 1310 1255 2478

Uniforms, school dress or other clothes and shoes

1025 1169 1271 1902

Tiffin 319 241 732 941

Informal payments to teachers or institution

43 1 0 0

Private tuition 657 1104 1532 5996

Other expenses 0 46 6 131

Total 3703 4920 5848 15462


A small amount of household spending is offset by support: many institutions provide in-kind support (94 per cent of primary non-madrassahs, 60 per cent of madrassahs, and 29 per cent of secondary non-madrassah institutions), while almost all secondary general stream institutions (97 per cent), 60 per cent of madrassahs and 61 per cent of primary non-madrassahs provide financial support. Much of this is accounted for by the government stipend schemes at primary and secondary.

3.4 Governance, management and community participation

On the basis of commonly used indicators of school governance and parental involvement, the registered system outstrips the newer entrants into the system such as NGO schools, kindergarten and hybrid schools, and many types of madrassahs in terms of performance. All schools reported having had visits from the relevant government, NGO or association officials in the past few years, and so there was no variation on this indicator. On paper, at least, registered schools were more likely to report having a functioning School Managing Committee than NGO and kindergarten schools, or than primary and non-formal madrassahs. Most of these school management committees (SMCs) in registered school types were also reported to have consulted with parents about school improvement issues over the previous year, and many to have raised funds or have provided other donations in kind.

A more mixed picture emerges with respect to the existence of Parent Teacher Associations (PTA). In theory, these should play an important role in enabling parents to participate in school management decisions of critical relevance to their children‟s education. On paper, government primary schools are most likely to report functioning PTAs; this contradicts other experience and other research which suggests that few schools have functioning PTAs. It

26

does suggest, however, that SMCs and PTAs are important elements of the formal structures and rule systems of government schools, and that they are more likely than not to be found in the official records. This highlights one of the significant problems with use of official school governance indicators to assess performance. By contrast, most NGO, community, kindergarten, hybrid schools and primary madrassahs reported not having PTAs. At secondary, dakhil madrassahs appeared to be the most likely institutional type to have formalised relationships with parents through a PTA: for most other types such an arrangement was rarely reported. Most schools do, however, hold events which parents can attend, and virtually all claimed to have systems through which parents can give feedback or complain about education service delivery. Just over half of schools (55) reported having a school development plan, but in only 12 of these were parents invited to be part of the planning process.

Table 3.10 Parent Teacher Associations

Institutions with Parent Teacher Associations: Yes No Number of institutions that held meetings in previous year

Primary Institutions

Government primary school 11 3 11

Registered non-government primary school 3 2 3

NGO primary school 1 9 1

Community primary school 1

Kindergarten school 1 2 1

Primary madrassah 3

Hybrid primary school 1

Secondary Institutions

Government secondary school 1

Registered non-government secondary school 13 18 13


2 4 2

Dakhil madrassah 5 4 5

Quomi secondary madrassah 1

Higher Institutions (but offer secondary and/or primary classes)

Registered non-government intermediate college 1

Alim madrassah 1 4 1

Fazil madrassah 1 2 1

Non-formal madrassahs

Non-formal madrassahs 9


In assessing school governance and parental involvement, it is worth noting that as types, NGO and kindergarten schools look worst on paper: across most indicators, they appear to have the least formalised systems for ensuring accountability and participation. Some types of madrassahs also appear to have limited means of engaging parents or answering to their needs. This issue is of considerable interest because of the expectation that school choice

27

might lead to improve quality provision, in part because competition will lead to stronger pressure for accountability at the school level. Yet qualitative evidence suggested a view that NGO and many of the private schools were more rather than less accountable, despite their lack of formal structures and systems. In the case of NGO schools, close supervision from the NGO was cited as important, but the fact that many NGO schoolteachers are local women is also likely to help poor parents communicate with school authorities. With kindergartens, the payment of fees and schools‟ dependence on retaining students are seen as creating incentives to perform. Many types of madrassahs are also rooted in communities and substantially dependent on their charitable and other forms of support; these connections, and their role in communities as spaces for the conduct of religious ceremonies and preaching through Awaaz Mahfil, milads, and celebration of religious occasions such as EId-i-Miladunnabi, all bring parents into closer contact with them. This may be one reason why teachers were less than half as likely to be absent on the day we visited the unregistered institutions in our sample, compared to in the registered institutions. The situation in government and registered non-government schools with respect to accountability and participation is more mixed, however, and the quality of school governance is likely to depend on the nature of the relationship between schools and communities, including the degree of local ownership over schools and the status and position of head teachers.

28

4 Aspirations and attitudes towards education

This section first presents findings about aspirations and expectations in relation to children‟s education, and then goes on to discuss how these shape and are shaped by educational preferences and attitudes towards education.

4.1 Education and attitudinal change

Some cross-cutting issues about attitudinal change in relation to education from the findings of the study are worth noting.

4.1.1 Education as the new norm

Across the study sites, there was wide acceptance that some degree of education is essential for all children, even among parents who have not themselves been formally schooled, or for whom there are no obvious economic returns to be gained. The level of education considered minimally necessary varies widely, with parents in the poor Kurigram community viewing basic literacy and some ability to keep accounts as adequate, while in the Gazipur community, where education was prized most highly among the four sites, parents felt that possession of a degree was the mark of an educated person. The finding of universal acceptance of at least basic education as necessary appeared to be rooted in two types of recent social change. First, a clear sense that overall levels of education have risen everywhere, with the decade of the 1990s commonly identified as the starting point for this expansion of education. This creates social pressure to invest in children. In a group discussion in Kushtia, women in their 20s and 30s noted that previously, there used to be one educated person in a village, but nowadays each household has educated people. People who were not themselves learned, did not see the point of schooling their children. But now they think, „everyone else is educating their children, so why should mine remain ignorant?‟ For people in their mid-30s and older, there is a clear difference between the experiences and attitudes of previous generations compared to the present context.

A second factor of interest provides empirical support to theoretical perspectives on educational investment. This is the sense that the decline in family size is associated with greater investment in children now than in the past. It is a view that supports Kabeer‟s argument that the demographic transition towards smaller family sizes contributes to household preferences for and capacity to invest in the „quality‟ of their children – as opposed to focusing on their quantity, in expectation of high infant mortality rates and dependence on children‟s labour (Kabeer 2001). A woman in Kushtia noted that:

If you have 10 children, and your income is from daily labour, how are you going to educate 10 children? Nowadays people have one or two children. So now they think, even if it’s very difficult to do so, they will educate their children. Because families are smaller, education has got better.

The finding that a basic minimal level of education for children constitutes an emergent norm is supported by two cases of households in which illness or disability were not considered grounds for failing to invest in children‟s education. In one Kurigram household, the ten year old daughter had a congenital condition which had resulted in skull disfigurement, and which was believed to result in early death. The parents explained they had every intention of educating her to the highest level possible; neighbours had encouraged them to do so

29

despite the sacrifice involved, because everyone felt she deserved every opportunity she could get. In one Kushtia household, although several of the children were mentally disabled, the parents invested in the education of each. The father said that:

For children like mine, it is necessary to have separate schools, so that they can learn too. You should tell the government to look into this. Because everyone needs education. These days, it’s not possible to neglect education.

4.1.2 Equivalent – but not equal - investments in girls and boys

Across the sites, the findings suggested that while household aspirations and expectations of girls compared to boys continue to differ, this did not necessarily entail stark inequalities in preferences for boys‟ as opposed to girls‟ education. Instead, there was a strong sense that there should be equivalence in how education is viewed and invested in for boys and girls, even though the strategies would not be identical. This equivalence was most marked at primary, after which gender distinctions become more relevant to household investment strategies. This finding suggests a significant degree of acceptance of the public discourse of gender equality, at least in education, which has been part of official strategies since before the 1990s through awareness-raising interventions such as the popular and enduring UNICEF „Meena‟ campaigns, as well as the strong government signals sent by the Female Secondary School Scholarship programmes (Hossain and Kabeer 2004).

People tend to voice views that they understand to be in line with public and official discourse, however, and there remain clear distinctions in aspirations and expectations for girls compared to boys, as well as in actual practices. While marriage is the expected norm for all, girls‟ marriage remains a particularly urgent issue for poor households, for whom successfully marrying off a daughter as soon as possible is a matter of considerable relief, as it removes the need to protect girls‟ reputations, reduces day-to-day living costs, and completes the costly lump-sum investment in dowry and marriage costs. The impacts of the pressure from the marriage of daughters are ambiguous, however. For many households, particularly those seeking upward social mobility, an educated daughter seems to be an asset, as she may make it possible for marriage into a more educated or higher status family. By contrast, poorer households, as well as wealthier households in the Kurigram community, tend to view the marriage of daughters as a problem to be resolved as soon as is feasible; older girls and more educated girls may also require larger dowries (see also Amin et al 2006). In one highly educated household in Kushtia, the household head‟s sister‟s failure to marry despite having passed her MA was a matter of concern and embarrassment, although it did not appear to limit that household‟s continued investment in its girls‟ education. Girls who have had madrassah education or even less formalised Arabic learning are also considered to have better marriage prospects, across the four communities.

30

Figure 4.1 What do you hope your children will learn or gain from their education? (% of households, unprompted responses)


Figure 4.1 presents unprompted responses from household heads and other adult respondents to the question: „what do you hope your girls or boys will learn or gain from their education?‟ The responses suggest that expectations of the purpose and outcomes of education do, indeed, differ for girls and boys, although only on a few issues were these differences substantial. While half of households felt that formal sector employment was a hoped for outcome of education for their boys, one-third felt the same was true for girls: although these figures need to be treated with caution as they are not based on a representative sample, this findings suggests a considerable shift in public attitudes towards women‟s employment, and a relatively high degree of optimism regarding the employment

prospects for educated women.12

More people expected girls would benefit from their

education through gaining academic skills, good marriage prospects, good domestic skills, and connections and friends to help find a job than that boys would do so. With respect to the impact on marriage prospects, 26 per cent of households suggested this was an expectation for girls, compared to only 5 per cent for boys. Other significant gender differences were on gains in terms of formal sector employment, self-employment and a reputation for knowledge and learning, all of which were more likely to be considered gains for boys than for girls.

12

This seems to reflect the fact that the formal sector employment prospects for educated women

have increased rapidly over the 2000s, leading to a significant narrowing of the gender gap at the upper levels of the formal sector labour market (Al-Samarrai 2008).

31

Aspiration and expectation with respect to educational investment appeared to diverge earliest and most strikingly between the genders in the poorest community. In Kurigram, the lack of fees, the stipend for poor children and a biscuit-tiffin programme all incentivise primary education, but an overriding motivation is that boys are unable to work before they are 12 or 13 and girls will not marry before a similar age. Therefore gender divergence strategies start just after primary, whereas for richer communities aspirations and expectations diverged later for girls and boys. In the Gazipur community, a class of highly educated professional appeared to have similar aspirations for their daughters as for their sons, possibly reflecting the rapid recent growth of professional employment opportunities for educated women at the top end of the labour market (Al-Samarrai 2008b). Less affluent households as well as the richest landowning class both viewed education up to at least SSC or Class 10 as the appropriate level for girls, at which age they were deemed marriageable.

A factor with strong potential to lead to gender-divergent strategies for educational investment was girls‟ safety (see Box 1, Security and School Choice). Women in Kushtia explained that there were „fears about girls‟, including that freedom to travel may also entail freedom to have boyfriends and love affairs. In Kurigram, too, it was felt to be better for girls to be schooled close to home.

4.2 Aspirations and expectations

4.2.1 Livelihood opportunities

Existing local livelihood opportunities provided the immediate frame of reference for parental aspirations with respect to children‟s education. In one view from the poor community of Kurigram,

Box 1 Security and School Choice

Children‟s security emerges as a factor influencing school choice, particularly for girls and younger children. A survey on human security in Bangladesh in 2007 found that education institutions topped the list of local institutions that people felt contributed to their security, suggesting that having schools close by made people feel safer about their children. This finding gains support from the School Choice study, in which distance to school was a key determinant of the school selection process.

Many parents have good reasons to be concerned about the security of their children. In the religious and culturally conservative Chittagong community, sending girls to schools at any distance from home caused great anxiety, because of harassment. One particularly pretty girl had dropped out in Class 7, among other reasons because she was regularly harassed and pursued by boys on her way to school; following this, her father forbade her to leave the house.

The Kushtia community has a history of insecurity and violence, but this has declined dramatically in the last four years, particularly since the Caretaker Government came to power in January 2007. But the area still suffers from insecurity and lawlessness, and after the recent kidnapping and rape of a neighbourhood girl on her way back from school, many parents had been reluctant to enrol their girls in secondary schools that were far away. Trafficking of women and children to India is an enduring concern in Kurigram, and there, too, parents made it clear that they felt safer if their children attended schools in the local area.

Sources: Saferworld 2008; School Choice study.

32

I can’t afford secondary school for my son, so what is the point of him studying now? And if he labours in the fields now, there will be immediate benefits from his income.

There was considerable consensus in Kurigram that some basic literacy and ability to keep accounts was enough education for a boy whose future lay in agricultural labour. For girls, that and some religious education were necessary requirements for marriage. Among the small number of large landowners who set the educational standard in the area, few proceeded beyond secondary education, and there was little enthusiasm. Most boys from rich households were understood to abandon education after secondary school in order to manage the property.

At the other end of the scale, formal sector employment opportunities within realistic range of Gazipur appeared to inform the view that investment in education, even if it involved sacrifices, made considerable economic sense. In one female-headed household in Gazipur, a widowed mother of three was struggling to keep her children in school, with some financial support from her overseas migrant brothers. All her children were in school and there was considerable emphasis on her daughter‟s education, stating that „I will educate my daughter up to metric (SSC, Class 10) level, no matter how difficult it is‟, with the explicit aim of enabling her to get a formal sector job, and be independent in future.

The impact of local livelihood opportunities was most diverse in Chittagong. The returns from formal education were not felt to be very high by the men in the area. The lack of industry in the area meant formal sector jobs in the local area were few. Even once educated, people return to farming the land, in what is a highly lucrative agricultural area. Others migrate, or pull rickshaws. Many study in madrassahs to become huzoor or alim; the concentration of madrassahs and strong demand for religious schooling in the area entails strong demand for teachers of Islam and Arabic. In addition, the educational requirements for migrant labour are not onerous: „even those who are educated, when they go abroad, drive cars‟, while those who are not educated do manual labour. Many men felt the need for technical and vocational education was greater for potential migrants than formal education. For those who seek coveted civil service positions, there was a view in Chittagong that it helped to graduate from the elite government secondary schools, rather than local ‟moffusil‟ institutions.

Existing livelihood opportunities did not totally constrain educational aspirations. Many people displayed awareness of alternative possibilities, and many, both adults and children, could cite examples of particularly shining role models for educational achievement, which made particular schools famous in the area (see Box 2, Role models and star performers). For some, the wider possibilities meant incentives to educate children beyond the requirement of their immediate options. In Kushtia, one girl had been enrolled in the nearby high school, to which she was able to travel with her neighbour, the head teacher. Having failed her SSC there, however, she moved to a second, more distant high school, where she passed both her SSC and her HSC (Higher Secondary Certificate, Class 12). She had acquired some sewing training during her education, and her father bought her a sewing machine after she completed her education. She ran a successful tailoring business for a time, after which she moved to Narayanganj, where a relative had secured a garments factory job for her.

33

4.2.2 Education and social status

The economic and employment context and the expectation of economic returns are not the sole determinants of aspirations, however. Enough examples exist of households aspiring to education against realistic expectations of economic returns to suggest that other factors also matter. Foremost among these is social status. One woman in Kurigram was strongly in favour of educating her children, an issue which caused great conflict with her husband and was implicated in his decision to remarry and leave. At issue appeared to the downward social mobility she had experienced as a result of her marriage: her own family had been large landowners, and her motivations appeared to include regaining that status. She said:

I am trying so hard. I had to steal crops and rice to sell to try to educate my oldest girl. I couldn’t manage it. To educate children you need both husband’s and wife’s support. I didn’t get my husband’s support … I wasn’t able to educate my sons. Only my (younger) daughter is still with me … even if I have to wear torn clothes I will educate my daughter, and one day won’t she be someone? It doesn’t affect me what people say ... I will have peace if I can fulfil my hopes for *my younger daughter]. When people hope, they can do much.

The impact of the desire for social status through education depends substantially on local social structure. In the predominantly extremely poor Kurigram community, the small number of richer households only needed to complete secondary school to retain their social superiority. By contrast, in Gazipur, SMC members argued that a degree was the minimum level necessary to be respected as an educated person in their community. In Chittagong, the desire to raise social standing meant rejecting government schools, where children

Box 2 Role models and star performers

Local schools gain reputation when they produce outstanding students. Role models for academic achievement are typically drawn from the professions. In Gazipur, a chartered accountant was described as the most respected citizen, because of his educational achievement. Many girls from the Gazipur village were respected for having gained formal employment, many as teachers, and some were even studying to be doctors - the pinnacle of socioeconomic ambition in Bangladesh. Many of these high-achieving girls were studying in Dhaka schools, however.

Rabindranath Tagore, whose Kutirbari is close to the Kushtia site, provided a significant cultural role model for the area. The strength of cultural, literary and spiritual tradition are used locally to explain why aspirations for education are relatively high, despite the context of insecurity and limited economic opportunity.

The most memorable role model was Rafiqul from Kushtia, whose story was recounted during a participatory school mapping exercise with local children and youth. Rafiqul is the son of a betel farmer. He had always been interested in science, even as a small child. One day, he read an article in Prothom Alo about an essay competition organized by

NASA, for children from poor families. The prize was to be a trip to space!

To everyone‟s amazement, Rafiqul‟s essay won first prize. He became very famous and he was even interviewed on the TV. His primary and secondary schools became famous as well. He also got a golden A+ in his SSC exam. Unfortunately, he could not claim his prize, as he was refused an American visa. But the schools he attended remain well known, and have become very popular in the area.

34

learned to speak in the local dialect, as opposed to more respectable „proper‟ Bangla. Within communities, people‟s aspirations were felt to be closely tied to the quality of school that households could afford to invest in sending their children to.

Livelihood and social status ambitions in relation to children‟s education are also shaped closely by parents‟ own educational attainments, and the livelihood and social status outcomes those result in. There are also instances, however, of households strategising for social mobility through educational investment, even where this involves considerable sacrifices (as in the case described above). The issue of parental educational attainment in shaping school choice is explored in more detail in Section 5 of this report.

4.3 Perceptions of schooling characteristics: quality, access and religiosity

What kinds of schools do parents and children prefer, and why? Having seen key features commonly associated with quality in the different schools in the areas, this section turns to a discussion of how these and other characteristics of schools are perceived by people in the communities themselves. It should be noted that what people prefer and what they actually choose are not identical: distance and cost are critical constraints affecting actual choices, as the next section discusses in detail.

4.3.1 Academic standards and teaching

Overall, the issues most household members reported liking about the schools of which they had actual experience was academic standards and the quality and nature of teaching and of the teachers themselves (Figure 4.2). Exam results were only occasionally explicitly mentioned as an indicator of academic standards, although from the qualitative research it is possible to infer that „standards‟ chiefly refers to examination performance. The Gazipur community was most focused on academic standards, which fits well with other findings about the higher aspirations of the Gazipur community, and the greater importance assigned to formal education to achieving those goals in that area. The precise elements of what constituted high as opposed to low academic standards were not entirely clear, other than with reference to examination results. Beliefs about the standard of education provided by an individual institution appeared to reflect a combination of reputation earned over time, possibly signalled by former pupils‟ achievements in later life, and by the characteristics of those who attend. The issue of discipline, rules and the overall academic environment also appeared to be intrinsic to understandings of standards. This is discussed in more detail below.

There was greater clarity around what was liked and disliked about teachers. Efficiency, meaning ability to impart lessons; sincerity, suggesting commitment to teaching; good behaviour, indicating using corporal punishment appropriately; caring about and monitoring students‟ progress; and having male or female teachers (depending on the gender of the respondent) were all identified as likeable qualities of teachers. In terms of teaching style and quality, insights from the focus group discussions suggested a preference for teaching styles that enable children to understand rather than memorise their lessons: in a focus group with women in Kurigram, the differences in teaching style between different types of school were clear:

In the government school, the students simply go and come back. The teachers give no importance to lesson time. Whether the children are learning properly or not they don’t pay any

35

attention. They talk amongst themselves. And here all the teachers are women. And women feel tender towards children that’s why they cannot discipline them or beat them. Even if men teachers can’t teach, at least they can put pressure on. Those who do well in the primary, they have to be taught at home. But the type of teaching is good at BRAC. There they learn with fun, sing, do physical exercise - this is the little [pre-primary] BRAC school. And in the big [primary] BRAC school, they first give the lesson verbally, then they write it on the board, and then you don’t have to study at home. In the government primary school there are lots of children studying altogether, its really noisy, can’t keep an eye. And in the BRAC school they only have 30 students, so they can monitor well. And in the Islami Foundation schools they write things on the blackboard and explain the Arabic meaning, then memorise it. And in the maktabs they don’t teach any meanings, they just teach them to memorise Arabic.

By contrast, „scolding‟ teachers, teachers who sleep, smoke or talk on their mobile phones during lesson time were all disliked, particularly by children. In a number of cases, dislikes include insufficient teachers, particularly for key subjects like English and Maths.

4.3.2 Access and physical facilities

Access and proximity are crucial attributes of schools. Ease of access ranked third of all the features that people liked and difficult access ranked third of all the features that people disliked about schools of which they had experience. Schools that were nearby, within the village itself, or to which roads and access were good were liked. Schools that were considered distant, or to which access was difficult for other reasons – in Chittagong because a school was at the top of a hill; in Kurigram, because flooding and riverbank erosion limit travel during the wet season – were disliked. Access was the single most-cited factor explaining the choice of school which children were currently attending, and in three of the four areas, it was the reason most frequently given.

The obvious advantages of proximity to the household include the lack of need for transport and associated costs, and travel time. Within the preference for schools that are close to home, in addition, is a considerable degree of concern about children‟s security. This emerges most strongly from the qualitative evidence, and in relation to girls and smaller children (see Box 1, above). It is worth noting that parents have good reasons for being concerned about their children‟s safety en route to school, particularly during the rainy season: drowning is the leading cause of death for children aged 1-9, and transport injuries, animal bites and electrocution - all potential hazards of a long school journey, are among the other leading killers for children aged 5-9 years (Rahman et al 2005).

In none of the four areas were school facilities important features of what people liked about schools. However, in all four, and particularly Kushtia, the poor quality of the school building and physical facilities ranked high among issues people reported disliking. Children felt particularly strongly about facilities, a matter of some significance given that many poorer children exercise considerable agency over decisions to attend school. The Kushtia community shared the view that school facilities were poor, with more than half of all responses describing various aspects of the quality of the building, furniture, toilets, drinking water, playground and so on as features of the schools they had attended that they had disliked. The complaint that „rain dripped through the roof‟ suggested that the physical quality of schools in that area was particularly poor. The complaint of „cracks in the school walls‟ also suggested a concern about safety. Another feature that linked safety and facilities that

36

emerged from the qualitative work was the issue of boundary walls and ID cards: few institutions had either, but those which had been singled out for their good security practices (see the Box 3 on children‟s views on what makes a good school). Although private and KG schools are preferred for their academic standards and teaching quality, many are housed in converted houses and other non-purpose built facilities, lacking space for outdoor play or other activity.

Box 3 Children’s views on what makes a good school

Children in Gazipur gave their views on what makes a good school during a participatory school mapping exercise. Children felt the best primary school in their area, the ‘Flower’ Child and Institute KG school was good because:

It is a private school, and private schools have incentives to learn

The teachers are good, sincere and competent

They have cultural events and educational school trips

There is a daily homework diary and the class work is completed daily

The environment is safe and everyone uses their ID cards

The school has a canteen and also a library where books can be borrowed

If there is an emergency and a child has to leave the school premises, they will be accompanied by a staff member.

Poor children cannot afford to enrol in the Flower Child and Institute KG school and many instead go to the local primary school, which is convenient because there are no tuition fees, textbooks, stationery, bags and until recently pencil cases and pencils were provided; and some children from very poor families get a stipend.

The most popular secondary high school is the „People‟s‟ High School, because:

Recent SSC results have been good, with students receiving A+

Teachers educate the students with care and attention

There is an atmosphere of discipline and order

Classes are regular

Teachers provide their students with a sense of security

Access is easy and the roads are good

Educational school trips are organised.

Other features of a good school according to children and youth in Gazipur were:

An SMC which makes decisions not on influence and power but on merit

The Principal has to be a strong personality and present during school hours

Students take responsibility for maintaining the school environment.

37

4.3.3 Discipline, governance and the school environment

Another set of issues which people raised was to do with the regime of rules and the quality of the school environment which resulted. People spoke approvingly of schools with discipline, where classes were held regularly, and where there was a „congenial‟ environment for study. Systems for enabling teachers to monitor and supervise student performance were also noted as positive features.

Discipline and the school environment seemed particularly important to people in Chittagong: there the focus was less on academic standards and examination results than on the quality of the learning environment and discipline for both teachers and students, which seemed to be most important to the Chittagong community. The enforcement of rules for teachers was a significant factor across the areas: unpunctual teachers, failure to discharge teaching duties, partiality towards some students, and other failures to regulate teacher behaviour were all widely reported as reasons for disliking schools.

Figure 4.2 Likes and dislikes about schools ever attended (% of responses)

Notes: Responses based on open-ended responses from parents and children. 2) Includes all schools that each child has ever attended. 3) source: household survey, School Choice study.

4.3.4 Religious schooling

Findings about why some people prefer religious schooling suggested a number of explanations. One that emerged from the madrassah staff themselves was that the change in their official recognition and status means that graduates of registered madrassahs now officially stand equal chances of entering general universities and the civil service. The disincentives for attending ebtedayee, dakhil, alim or fazil madrassahs within the officially registered and supported aliya stream have largely declined, although some discrimination is believed to persist. In some households and particularly among madrassah students

38

(including at least one girl), religious education preferences were explained with reference to ambitions to become religious teachers. This makes considerable sense in contexts such as Chittagong where the growth of the madrassah sub-sector in itself has entailed a rising demand for madrassah teachers. In some households, having relatives who migrated to the Middle East appeared to create a preference for religious education, partly because it provides a form of social distinction. But some also believed that having Arabic language skills increased their chances of successful overseas migration.

The most common explanations of preferences for religious schooling included the desire for religious blessings and social respect associated with religious learning. For girls, this was seen as a means of enhancing marriage prospects. In a number of households, children were placed in madrassahs to fulfil the dying wish of a family member, or to honour a religious promise or prayer. Also common was the highly practical explanation that in aliya madrassahs, children are able to learn Bangla and all the other subjects as well as Arabic and religious studies; it was seen as a distinct advantage to provide all of these lessons in one site, rather than having to go to the maktab or an Arabic teacher in addition to general stream school. And finally, madrassahs are also seen as places to send boys in need of discipline and a firm hand, which the general schools are seen as somewhat less able to provide.

4.3.5 Other factors

Other reasons people prefer schools include the provision of religious or Arabic teaching, which made up seven per cent of all responses from the Chittagong community, and reflected the higher provision of religious-based schools in that area. Single-sex (girls‟) school status were another positive feature noted on a number of occasions. Other factors that people disliked included being harassed by boys in the area, not receiving the stipend despite being poor, and, in one student‟s case, other children refusing to sit next to him because he was poor.

It is worth noting that cost and the provision of material support featured very little in assessments of what was liked or disliked about schools: one explanation of this is that cost and the availability of financial support are the prior and perhaps most crucial determinants of school choice. Prior knowledge of the cost and material support features of schools provide the basis for decisions about whether to attend or not; these are therefore unlikely to feature prominently in people‟s assessments of institutions actually attended.

39

5 Schooling choice and outcomes

In this section, actual patterns of schooling choice and its effect on outcomes in the four communities are explored. This section will tie together the previous two chapters which examined school supply and household aspirations, in an attempt to shed light on why households are choosing to send their children to certain types of schools and the probable effect of these decisions on future livelihood choices and socio-political attitudes.

5.1 Actual schooling choices

Figure 5.1 presents responses to questions about the reasons for selecting schools currently being attended. Issues that appear to have been foremost in people‟s minds were practical matters of access and the quality of education. These reasons are most likely to reflect tangible factors or issues that were explicitly discussed during the decision-taking process; they are unlikely to reflect the influences and structural constraints that mean that some choices fall off the agenda before the full range of options could be considered. Some of these structural constraints are obvious: madrassahs were never an option for the Hindu and Christian households in our communities, for instance. Nor, however, were they in general likely to be seriously considered by wealthier households, unless they held unusually strong opinions about religion (which was the case in some households). Another constraint on the freedom of choice which may not be articulated clearly includes constructions of gender, including concerns about girls‟ security. Despite the recent rapid improvements in women‟s and girls‟ mobility in Bangladesh, limitations on their movement persist, even if these are not explicitly subject to formal rules or recognised as such.

Figure 5.1 Reasons for selecting current school


40

5.1.1 Poverty status and school choice

Household poverty remains a constraint on education enrolment in Bangladesh, despite initiatives such as stipends intended to reduce the opportunity costs of educational participation. In the present sample, 11 per cent of household members report having left school because of financial constraints; two-thirds of the 24 „never enrolled or dropout‟ households fall within the poorest quartile of the sample; and the largest proportion of students studying at secondary and higher level institutions are from the richest quartile.

The effect of poverty levels on the type of school chosen is less conclusive. Table 5.1 presents findings on the number of children attending each type of education institution according to poverty quartile. At primary, government primary schools attract the largest proportion of students from all poverty quartiles, and there are few clear differences in types of school across the four quartiles. At secondary, differences by poverty level are more apparent: children from wealthier households are more likely to attend non-government secondary schools and from poorer households to attend dakhil madrassahs. In general, madrassahs of all types attract a larger proportion of children from poorer households. It is important to highlight that the large number of observations for the richest quartile mainly captures greater school attendance, rather than larger household size.

Table 5.1 Educational institutions ever attended by children according to poverty quartiles

Educational Institutions

Poorest quartile

2nd quartile

3rd quartile

Richest quartile

Primary Institutions:

Government primary school 63 62 60 80

Registered non-government primary school 3 7 5 4

NGO primary school 9 16 19 8

Community primary school 3 0 0 4

Kindergarten 0 2 8 1

Primary madrassah 4 4 2 2

Secondary Institutions:


11 23 38 74


6 5 6 16

Dakhil madrassah 6 9 13 2

Higher Institutions:

Registered non-govt intermediate college 0 0 0 7

Government degree college 0 0 0 1

Registered non-govt degree college 0 1 0 8

Alim madrassah 1 1 0 2

Fazil madrassah 0 2 0 5

Non-formal madrassahs:

Non-formal madrassah 10 11 7 4

Notes: 1) The sample includes sons and daughters of the head of the household. 2) Includes all schools that each child has ever attended. 3) source: household survey, School Choice study.

41

5.1.2 Parental education matrices

As would be expected, parental education has an important effect on the education choices they make for their children. The level of education attained by fathers and mothers is positively associated with levels of children‟s schooling. There is less evidence about how the type of school selected for children is influenced by i) the level of parents‟ schooling and ii) the type of school attended by parents. The matrices below examine these relationships in the four communities, and demonstrate clearly that the depth and breadth of educational opportunities available to parents in our households was far smaller than is the case for their children.

Table 5.2 shows that educational opportunities have expanded considerably within one generation. The fathers of some 61 per cent of all the children in the sample had never enrolled in education. While the sample was purposively selected and therefore not statistically representative of the population in general, it is instructive to note that only 14 per cent of children in the present sample had never attended educational institutions; this is despite the fact that one quarter of all the households sampled were selected on the basis that they were likely to feature patterns of dropout or non-enrolment. Although most fathers who had had some level of schooling had continued to secondary education, only 12 children had fathers who had completed HSC to Bachelors pass level, and only two had fathers who had achieved education higher than that. Most children who had never enrolled also had fathers who had never enrolled.

Government primary schools (GPS) are by far the most common school type for children and the popularity of GPS is positively associated with fathers‟ education levels. Higher levels of father‟s education is also associated with a preference for kindergarten schools, suggesting that higher educated parents opt for schools which are widely believed to offer a higher quality of education. By contrast, NGO schools are particularly common among children whose fathers have never enrolled, supporting the idea that NGOs are targeting their services to families on the margins of educational services. Finally, children with less educated fathers are somewhat more likely to attend madrassahs than children with more educated fathers.

Table 5.2 Fathers’ educational level and children’s primary school types

Children’s school types

Fa

the

rs’

ed

uc

ati

on

level

Never enrolled

GPS RNGPS NGO Community school

Kindergarten/ English Medium

Ebtedayee madrassah

Quomi madrassah

Total

Never enrolled

50 139 11 41 7 4 6 1 259

Primary (grades

1-5)

5 43 5 2 0 2 3 2 62

Secondary (grades

6-10)

3 65 8 6 0 4 2 2 90

HSC-Bachelor

0 4 6 1 0 1 0 0 12

Higher 0 0 0 1 0 0 0 1 2

Total 58 251 30 51 7 11 11 6 425

Notes: 1) Students attending Dakhil madrassahs who have never been to any other type of school have been included in the Ebtedayee category. 2) Students attending RNGJS who have never been to any other type of school have been included in the RNGPS. 3) Other primary madrassahs and the hybrid school type have been excluded.. 4) source: household survey, School Choice study.

42

Similar patterns between parental educational level and school type for children are evident for mothers. Although, the proportion of children with mothers who have never enrolled is only slightly different to that of fathers, children are much less likely to have mothers who have studied beyond primary level. The mothers of most children in primary institutions never enrolled, and as is the case with fathers, children with less educated mothers are more likely to never enrol, attend NGO schools and attend madrassahs; similarly, children with more educated mothers are more likely to attend GPS and kindergarten schools.

Table 5.3 Mothers’ educational level and children’s primary school types


Mo

the

rs’ ed

uc

ati

on

al le

vel

Never enrolled



Ebtedayee madrassah

Quomi madrassah

Total

Never enrolled

48 138 12 35 3 2 7 3 248

Primary (grades

1-5)

11 83 5 11 0 4 1 1 116

Secondary (grades

6-10)

0 43 2 5 0 5 3 2 60

Total 59 264 19 51 3 11 11 6 424

Notes: 1) Students attending Dakhil madrassahs who have never been to any other type of school have been included in the Ebtedayee category. 2) Students attending RNGJS who have never been to any other type of school have been included in the RNGPS. 3) Other primary madrassahs and the hybrid school type have been excluded. 4) source: household survey, School Choice study.

Similar matrices were tabulated for secondary education. As with primary education, children are more likely to attend secondary institutions than their parents. However, the range of secondary schools attended by members of the communities is smaller than for primary. Children with more educated mothers and fathers were more likely to attend non-government secondary schools (registered), and children with less educated fathers to attend non-government junior secondary school (registered) and Dakhil madrassahs. None of the children in our sample attended government secondary schools and as a result we are unable to assess choice decisions made in relation to this school.

Table 5.4 and Table 5.5 examine the relationship between type of schools attended by parents and children, to explore whether parents who are exposed to a certain type of schooling are more likely to send their children to the same type of school. The findings are inconclusive, partly because the range of educational choices available to parents was limited, and partly because of the dominance of GPS in the system. Overall, the picture that emerges from is most likely to reflect the effects of differences in school supply by area.

43

Table 5.4 Fathers’ school types and children’s primary school types


Fa

the

rs’ s

ch

oo

l t

yp

e

Never enrolled



Ebtedayee madrassah

Quomi madrassah

Total

Never enrolled

50 139 11 41 7 4 6 1 259

GPS 8 93 19 9 0 7 5 2 143

RNGPS 0 14 0 0 0 0 0 0 14

Quomi madrassah

0 2 0 1 0 0 0 3 6

Don‟t know 0 3 0 0 0 0 0 0 3

Total 58 251 30 51 7 11 11 6 425

Notes: 1) Students attending Dakhil madrassahs who have never been to any other type of school have been included in the Ebtedayee category. 2) Students attending RNGJS who have never been to any other type of school have been included in the RNGPS. 3) Other primary madrassahs and the hybrid school type have been excluded. 4) source: household survey, School Choice study.

Table 5.5 Mothers’ school types and children’s primary school types


Mo

the

rs’ s

ch

oo

l t

yp

e

Never enrolled



Ebtedayee madrassah

Quomi madrassah

Total

Never enrolled

48 138 12 35 3 2 7 3 248

GPS 9 101 15 14 0 5 5 3 152

RNGPS 0 16 5 4 0 8 0 0 33

NGO 0 2 0 0 0 0 0 0 2

Ebtedayee madrassah

1 7 0 0 0 0 0 0 8

Don‟t know 1 7 0 0 0 0 0 0 8

Total 59 271 32 53 3 15 12 6 451

Notes: 1) Students attending Dakhil madrassahs who have never been to any other type of school have been included in the Ebtedayee category. 2) Students attending RNGJS who have never been to any other type of school have been included in the RNGPS. 3) Other primary madrassahs and the hybrid school type have been excluded. 4) Mothers attending different types of primary schools accounts for the difference between Table 5.5 and Table 5.3. 4) source: household survey, School Choice study.

5.1.3 Gender and school type

Participation rates were similar among all the children sampled, with 85 per cent of boys and 88 per cent of girls currently or previously enrolled. By contrast, most of their parents had never enrolled. That participation rates were marginally higher for girls than boys also represents significant recent social change. In Gazipur, people noted that government policies to encourage female secondary school attendance had stimulated demand for primary schooling among families who had not previously considered sending their girls to school, encouraged by the financial support available.

The distribution across school types is similar by gender, particularly at primary. Government primary schools are by far the most popular type of primary school, attended by 77 per cent

44

of boys and 67 per cent of girls (Figure 5.2, Figure 5.3). The main gender differences are in NGO schools and registered non-government primary schools (which proportionally more girls in the sample attended) and unregistered madrassahs (proportionally more boys). NGO schools, and particularly BRAC schools, are known to focus on attracting girls into school. It is less clear why more girls than boys should be attending registered non-government primary schools; one possibility is that distance is a determinant in this case, as parents may view proximity to be more important for girls than for boys, both for considerations of security and for time taken from domestic tasks which primary aged girls are more likely to engage in than boys. Second, in line with common perceptions of unregistered madrassahs, boys in our sample were more likely to attend such institutions than girls, a pattern that continues through all levels of this sub-sector. However, it is interesting to note a slight gender difference in favour of girls in unattached ebtedayee madrassahs at primary, a difference which continues into secondary within the registered madrassah sub-sector.

Figure 5.2 Primary school types attended by male children of households

Community

school

1.5%

Kindergarten/

English

Medium school

2.9%

Ebtedayee

madrasah (not

attached)

1.0%

Government

primary school

76.6%

Quomi

madrasah

(unregistered)

2.4%NGO school

(primary)

11.7%

Registered non-

government

primary school

3.9%


45

Figure 5.3 Primary school types attended by female children of households

Community

school

2.5%

Kindergarten/

English

Medium

school

3.1%

Ebtedayee

madrasah (not

attached)

2.5%

Government

primary

school

67.1%

NGO school

(primary)

17.4%

Quomi

madrasah

(unregistered)

0.6%

Registered

non-

government

primary

school

6.8%


At secondary, gender differences begin to emerge; again, this finding is in line with findings about aspirations and preferences around gender and education as discussed above. Both boys and girls are most likely to attend non-government secondary schools, but the likelihood is greater for boys. Girls are more likely than boys to attend Dakhil madrassahs and non-government junior secondary schools. The different choices made regarding secular and non-secular education according to gender may be explained by future livelihood expectations. As noted above, the view that girls with madrassah education enjoy better marriage prospects is fairly widespread among Muslims. And despite some evidence of rising aspirations for women‟s employment, parents continue to invest their employment ambitions and hopes in their sons; in some views, formal sector employment was more likely if children had attended registered general stream schools. This seems to be a hangover from the period before aliya madrassah education was accepted as equivalent to general stream education.

5.2 Schooling choice patterns within households: stylised facts analysis

From the combination of the quantitative and qualitative evidence on how schooling choices come about, a number of stylised facts about schooling choice patterns can be identified.

First, wealthier households with a higher stock of parental education are more likely to have access to and to „shop‟ between different types of school. This reflects the following related factors:

46

They are more likely to live in relatively affluent regions, where school supply is relatively diverse and schools are more numerous;

Employment opportunities are more likely to be of higher quality, require higher levels of education, and therefore to require more educational investment;

The parents are better-positioned financially to consider and select between different schools;

The parents are likely to have more information about and experience of schools themselves, and therefore to evaluate school quality and performance.

The schooling choices of the Haque family in Gazipur highlight these stylised facts about the economic and normative impact of area in combination with wealth and the stock of education. The Haques are a small, relatively well-educated and relatively wealthy family in Gazipur. The father has a BA degree pass qualification, and is self-employed, and runs a training centre, which provides the main source of income for the household. The mother studied up to SSC level, and both parents studied in government primary and non-government secondary schools, while the father completed his education in a government degree college. Both of their children, a 17-year old boy and a girl of 11, are currently studying; they have attended a variety of institutions. Their son, having attended a government primary school now attends a non-government secondary school. The family approve of his current school because it has high academic standards and modern teaching methods, as well as offering study tours and a quiet and peaceful setting. This may be important because his mother recalls disliking her secondary school because it was set just beside a noisy busy road. But there are problems with the school: it is beside the bazaar, and is sometimes short of teachers. Overall, the school was chosen mainly because it is quick and easy to get to. The 11 year old daughter never entered the government system, and went straight to a kindergarten/English medium school. Again, this choice was reported to be mainly because of ease of access. But it is also liked because it is has high academic standards, is in a peaceful and quiet setting, and the teaching methods are considered modern. The main problem with it is that, in common with many kindergarten schools, the building is of poor quality: in this instance it is made of tin, and gets very hot in the summer. Both children are expected to study up until Masters level.

Second, capacity to pay does not automatically entail that people will be highly selective, nor that they will choose high levels of educational attainment: other factors include community norms around education, the extent to which economic opportunity depends on educational attainment, and the stock of family education. The Nur Mohammed family‟s schooling choices reflect such a set of stylised facts. They are a large and wealthy family in Chittagong: the fact that they have seven children – five sons and two daughters – suggests they may be less likely to view investment in children‟s education as a high priority, consistent with other more traditional views on family and the value of children. The father has been self-employed in the lucrative but illegal timber trade, which has been more or less stopped under the Caretaker Government crackdown on illegal activity. He is a member of the influential local elite, as a former Union Parishad chairman and member of the informal dispute resolution system, and linked to the mosque committee and activity. The household‟s main source of income is from remittances, from the one son employed abroad, and there are plans to send another soon. Three other sons are self-employed in the profitable agricultural produce trade. Both parents attended school, with the father studying up until class 7 and his wife up until class 3; he left because that was the highest class the school went up to (he is 63) and his wife left because in those days religious practice prohibited older girls from studying higher (she is around 50). All of their children have completed at least primary education. The daughters left school before their brothers, and are both now married and living in their married homes. Like their mother, neither have a formal occupation in addition to their

47

domestic work. Most of the sons left school of their own accord during secondary education, and started working. All their children went to the same government primary school, which they view as „our school‟, because it is an old local institution; they all also attended the same non-government secondary school because it was easy or quick to get to. Views on the schools they had attended did not show signs of deep consideration of school quality, and most explained they liked the high standards but disliked the inefficient teachers of the schools they had attended. When we visited that household, the youngest son was just considering withdrawing from secondary school, because he wanted to go to work abroad.

Third, poverty and vulnerability remain the critical and most widespread constraints on household capacities to choose between school types. Very poor households with limited assets in remote areas with relatively few economic opportunities often have limited aspiration for educational investment: the rational and most likely outcome under these conditions is that children with sufficient aptitude are likely to attend primary for some period, so long as these are accessible, of reasonable quality, and offer some incentives (cash, food or other material support). Periods of household or wider crisis or shocks such as death of family members or riverbank erosion episodes, or if, as was the case in the poorest community, a popular school closes down, children may be permanently withdrawn. While choice is limited in such circumstances, poor households may choose between different types of low investment strategies, such as attending either the government primary school or the local maktab, neither of which cost much in terms of time or cash, but which yield some returns in terms of useful literacy and numeracy, earning religious blessings and social respect, and for girls, possible improvements in marriage prospects.

The Fazal Haq family in Kurigram highlights some of these stylised facts about how poverty shapes school choice. Theirs is a large family, in the traditional north Bengal manner, with seven children – five boys and two girls. Both parents never had the chance to enrol in school, and the father is a daily wage labourer, while his wife manages household domestic activity. Their oldest son, now 23, was the first in the family to enrol; he studied up to class 5, at which point he left education. None of his brothers aged 19, 13 and 10 ever enrolled in school, and his only other sibling to try school is his 16 year old sister, who lasted until Class 3, at which point she had to withdraw for financial reasons. Both attended the local government primary school. It should be borne in mind that job prospects for men are limited in the Kurigram community and surrounding area, and more so for women. Boys are widely expected to – and indeed most do – enter into agricultural wage labour with some small agricultural produce trading on the side – as soon as they are physically able, at some time after primary school going age. Girls marry as soon as their parents can identify a suitable husband and save up the dowry. The pressures and attractions of dowry mean that both boys and girls are married very young in the area, no doubt contributing to the high birth rates, and the lack of a demographic transition to smaller families with a higher predisposition to invest in education. Vitally, the area is highly environmentally insecure, and the local government primary school has had its location shifted four times due to flooding and the shifting of the riverbanks. Land ownership, agricultural activity and livelihood security are all affected by these conditions; educational investments fall low on the list of household priorities under these conditions.

Fourth, poor households in poor areas are more constrained in their schooling choice than poor households in more prosperous areas, where economic opportunity is greater, and choice wider. An example of this fact is in the schooling decisions of Nasima Begum, the young widow head of a relatively poor household in Chittagong. Nasima Begum has 3 children; 2 daughters and 1 son. Although she never enrolled in school her children are all currently studying. Two of her daughters are attending a non-government secondary school, which they attend because it easy/quick to get to. Their brother is in the primary cycle and

48

after attending a non-government primary school until class 3 is now attending an Alim madrassah, which Mrs Nasima Begum chose because she felt her son would be safe going there.

5.3 Outcomes of school decisions

5.3.1 Labour market and employment

Higher status employment, such as in the formal sector, were more common among household members with higher levels of education, while those with lower levels of education were more likely to be daily wage labourers or no fixed occupation. Self-employment made up a high proportion of outcomes at all levels of education, however this is largely because it captures a wide range of income and wealth employment opportunities available to households, dependent on the type of self-employment. A small number of women with higher levels of education were also working in the formal sector, but by far the most common occupation across all levels of education up to grade 10 for women was unpaid domestic work for own household. Yet in a third of the households, it was indicated that their women would be prepared to work outside the home if jobs were available to them (Table A.4 and Table A.5).

In terms of aiming for further and higher education as a route into more lucrative, exclusive labour markets, expectations of those household members that were currently studying varied between areas. Students in Kurigram had lower expectations about their educational attainment, where some students expected to achieve either a maximum of grade 5 or grade 8, while 60 per cent of all students expected to complete only a maximum of grade 8. This is illustrative of the limited enthusiasm for education expressed by households and supports the findings that a large proportion of households leave education to manage property, with small numbers progressing to secondary. As expected, higher levels of educational aspirations were expressed by households in Gazipur, with 50 per cent of students expecting to achieve degree level or higher qualifications. The higher economic opportunities in Gazipur and the proximity to economic opportunities in Dhaka seem to stimulate the demand for higher levels of achievement. The same levels of educational aspirations were also reflected in Chittagong, however this may be a result of the higher proportion of wealthy households sampled in this area and may reflect the lucrative agriculture opportunities available to individuals. The results in Kushtia appear to reflect that community‟s unusually strong preference for education: 47 per cent of students there expected to achieve degree level or higher qualifications, and no current students expected to achieve below grade 8.

Analysing the associations between employment opportunities and schooling type, for those not currently studying, is shaped by the smaller school type choice set that was available to older household members. As mentioned above, it is clear that a generation ago the range of school choice was significantly smaller than is the case today. Not only did the majority of individuals not enrol in education, but the range of schooling types available to them was significantly smaller than that available currently in Bangladesh. For male household members that attended primary school, over 90 per cent attended government primary schools, and at secondary, most attended registered non-government schools. For women, the attendance at different institutions is slightly more diverse.

The limited choice of different school types available to older household members limits analysis of the links between institution types attended and future employment. The increased choice of schooling type, particularly at primary level, but also at secondary level, is a relatively new phenomenon in Bangladesh. Studies that take place in a few years that capture the current occupation of learners that have attended the wide range of institutions

49

currently available will be able to more usefully assess the relationship between different school types and future livelihoods. This will be particularly interesting for the women, who in a generation have faced the largest shift in educational opportunities. Despite the very narrow livelihood opportunities currently available to women, over half of households sampled hoped that education would create employment opportunities for their female children.

Discussions with community members suggest that some people believe government and registered secular schooling provides children with greater economic opportunities than non-secular schooling. However, some more educated families were aware that following the incorporation of Dakhil madrassahs in the government education system curriculum and standards were improving and it was felt that economic opportunities from studying at madrassahs would improve. Some households felt that religious schooling was for less able children. While others felt it was only useful for children who wanted to work within the religious sector, which would perhaps explain the popularity of madrassahs in Chittagong.

5.3.2 Social and political attitudes

Changing preferences for education levels and school types have potentially profound

implications for social and political attitudes.13

The issue of how education affects social and

political attitudes takes on particular urgency in present-day Bangladesh in the context of a) concerns about the potential for radical political Islamic attitudes to be generated through madrassah education; and b) an apparent rise in enrolment in religious education institutions of a range of types, reflecting in turn the possibility of a shift in preferences towards religious away from general education.

This section of the report explores the study‟s findings on secondary school students‟ attitudes towards social and political issues, including attitudes towards democracy, minority group and human rights, and gender equality. Data on these attitudes were collected through a Civic Education questionnaire administered to Class 9 (or equivalent) students in all secondary institutions, both general stream and madrassahs of all types. The questionnaire was adapted from authoritative instruments used to collect political attitudinal data (the World Values survey and the regional Barometer series - a recent cross-regional survey of attitudes towards democracy in South Asia). The full instrument is available in the annex. Students filled in the questionnaires themselves, after a brief explanation of how to do so. While there were a significant number of non-responses to some questions, careful pre-testing of the instrument meant that in most cases, most students understood the questions, even where difficult formal language was used out of necessity. It is worth noting that in Bangladesh, a sample of class 9 students is likely to have above average levels of education; this entails that comprehension of questions should be less of a concern with respect to the data collected than is the case in some cross-national attitudinal surveys (e.g. the World Values Survey) of which Bangladesh is part.

The purpose of exploring political and social attitudes of students was to test the following questions:

1. Are students from religious institutions less supportive of democracy and more supportive of non-democratic alternatives than students in general stream education?

13

Based on analysis of attitudinal data from 18 African countries, Evans and Rose (2007) find that education is a far more important influence on democratic attitudes than other factors associated with modernity, such as urbanisation, economic status and religion.

50

2. Do students from religious institutions support minority and human rights and gender equality principles to the same extent as students in general education?

3. How does gender affect social and political attitudes?

4. How do other differences in school type affect attitudes?

5. Does regional local affect social and political attitudes?

This component of the study was motivated in part by recent evidence about the impact of religious education on social and political attitudes in Bangladesh collected by Asadullah and Chaudhury (2006). Based on a small sample survey of female graduates of general and government-registered (aliya) madrassah institutions, their findings were that madrassah education was consistent with pro-democratic attitudes, but also with more conservative social attitudes, for instance towards fertility control and women‟s employment and education. The Civic Education survey component of the School Choice study aimed to explore these issues in some more detail. The sample covered 1083 girls (150 in madrassahs and 933 in non-madrassah institutions) and 700 boys (133 in madrassahs and 567 in non-madrassah institutions), a total of 1783. These covered all children in class 9 or equivalent in all secondary institutions within our four catchment areas in which it was possible to administer the questionnaire.

Table 5.6 presents findings of the class 9 students‟ preferences for different forms of

government.14

Multiple preferences were allowed in the survey, and respondents were not

required to choose between the different types of government. The responses suggest an overall strong support for government by a leader elected by the people through a fair election, with 88 per cent overall supporting this form; madrassah students were slightly less likely to support this than those in the general stream, but still overwhelmingly agreed or strongly agreed with this form of rule, at 85 per cent of all madrassah student responses. Rule by experts came close second overall, possibly a reflection of a strong degree of support for the then-Caretaker Government, led by experts. Just under two-thirds of the students supported rule by the Army, with girls somewhat less likely to be supportive than boys. Interestingly, the idea of rule by a king, or inherited rule, was slightly more popular than the idea of rule by religious leaders. However, madrassah students were significantly more likely to support the idea of rule by a king (62 per cent compared to 45 per cent in the general stream), and even more likely to support government by religious leaders (73 per cent compared to 40 per cent). Although girls displayed a slightly stronger disposition against non-democratic modes of rule (slightly less support for rule by experts, the army and religious leaders) than boys, 6 per cent more girls supported an authoritarian ruler than boys. Notably, while only 29 per cent of madrassah students supported authoritarian rule, this was some 10 percentage points higher than non-madrassah students. The findings from this survey question suggest that while a preference for democratic rule is overall widely held, there is also a high degree of tolerance for non-democratic forms of governance that the secondary school students are likely to be familiar with or to have experienced, and that ideas about alternatives to democracy which feature a prominent role for inherited and religious leaders are particularly attractive to madrassah students.

14

The responses were to the following question: „There are different ways in which a country may be

governed. What do you think about the following ways? a) We need an authoritarian leader who does not bother about elections; b) The country should be governed by the Army; c) The country should be governed by a King; d) The country should be governed by those chosen by the people in a fair election; e) All major decisions about the country should be taken by experts rather than politicians; f) All major decisions about the country should be taken by religious leaders rather than politicians.

51

Table 5.6 Preferences for different forms of government

% strongly agreeing or agreeing with: Boy Girls Madrassah Non-madrassah

All Frequency

Elected leader 89 88 85 89 88 1,482

Experts 83 82 81 82 82 1,323

Army 64 61 63 62 62 990

King 47 47 62 45 47 774

Religious leaders 46 43 73 40 44 716

Authoritarian leader 17 23 29 19 21 328

Notes: 1) source: civic education survey, School Choice study.

Beliefs about democracy were consistent across gender and school category, however, as shown by Figure 5.4 across the groups there was almost no concern about the protection of minority rights, vulnerable groups, or domination by the majority. The overwhelming majority liked the freedoms associated with democracy, and more than half felt that a rise in corruption associated with democracy was its worst feature. Although these findings do not demonstrate significant differences across school type, they are still of interest in that they highlight how closely future citizens associate democracy with corruption, as well as their lack of awareness about the role of democracy in the protection of the rights of marginal and vulnerable groups.

Figure 5.4 Beliefs about democracy (% responding): features of democracy liked most


52

Figure 5.5 Beliefs about democracy (% responding): features of democracy liked least


Figure 5.6 Views on the treatment of citizens (% agreeing or strongly

agreeing)


Views on citizens‟ rights were also broadly consistent across gender and school types (Figure 5.6). The issue on which the highest proportion were in agreement or strong agreement was that of higher education for girls. The slightly higher proportions of boys than girls and of madrassah compared to non-madrassah students that agreed with this proposition may not be significant enough to be surprising. The only substantial difference of opinion appeared to be on the issue of women‟s freedom to work, of which girls were more supportive of boys, and non-madrassah students than madrassah students. Looked at more closely, some 62 per cent of boy and 25 per cent of girl madrassah students who responded to the question strongly disagreed or disagreed with women‟s freedom to work; this contrasted with the 86 per cent of girl and 77 per cent of boy non-madrassah students who responded by agreeing or strongly agreeing that women should have that freedom.

The findings from the Civic Education survey component of the study are indicative of areas in which madrassah education is yielding shared and progressive attitudes towards democracy and rights, as well as highlighting specific issues on which there appears to be a considerable gap in social attitudes.

53

6 Discussion and conclusions

This report has presented findings from a study of school choice in Bangladesh, focused on three sets of findings:

a. School supply; the numbers and diversity of school types; the quality of educational inputs, physical infrastructure, governance and management, level of educational attainment, cost and access

b. Preferences and aspirations with respect to types and levels of education, including attitudes to education, aspirations and expected returns from educational investments, and perceptions of school quality; and

c. Patterns of actual school selection and (to a more limited extent) their socioeconomic and socio-political outcomes.

The study was based on in-depth comparative case studies of households, communities and schools serving four rural areas in Bangladesh. This section of the report analyses the findings in greater depth, and discusses some of the key implications for policy.

6.1 School supply

The headline finding with respect to school supply is that in terms of the numbers and diversity of educational institutions that can potentially serve the communities in question, there has been a rapid recent growth. Much of this growth dates to the early 1990s, but a more recent wave of new institutions can be dated to the early 2000s. Newer institutions include a) NGO schools, most of which in the present sample are BRAC schools; b) a range of different types of institutions delivering religious education, some alongside the general stream of education and registered by the official system, but many also providing primarily Arabic language and religious instruction; and c) kindergarten schools, which are characterised chiefly by the fact that they are relatively costly private institutions mainly attended by wealthier children, which place considerable emphasis on discipline, time-on-task, communication with parents, and homework. „Hybrid‟ types of school, combining the advantages of close attention to educational inputs of private kindergarten schools with some religious instruction were also identified as a new and highly popular category of institution. The present study is one of the first to have conducted any in-depth research into unregistered educational institutions in Bangladesh; the disadvantage of this is that the necessarily small sample size entails that there is no means of assessing whether these findings are generalisable on any more significant scale. In terms of trends, however, the growth of new and more diverse institutions suggests an overall widening of school supply, and seems likely to reflect the growing demand for different types of institution.

An important finding of the study is the lack of any clear direction with respect to how wider choice impacts on the quality of schooling on offer. Globally, debates on school choice have revolved around whether or not wider choice leads to higher quality, with proponents of greater choice arguing that it does. The present study findings, by contrast, show no clear patterns in support of such a claim. If anything, schools which serve poorer households in communities with greater school choice appear to have lower educational attainment than comparable schools in areas with less choice. These findings are consistent with the claims of critics of school choice policies that wider choice can lead to „cream-skimming‟ and rising inequality. The findings here are suggestive, but not conclusive, of a process of wealthier households withdrawing their children from the (generally poorer quality) government and registered non-government schools at primary (a comparable range of private-public

54

institutions are not available at secondary). The findings are indicative only, and suggest a need for attention to this important issue of quality and inequality in educational access.

The evidence on quality is mixed, yet in terms of physical facilities, government and other registered institutions are clearly far better endowed than the newer types of and unregistered institutions. Apart from when they are extremely poor or even dangerous because of cracked walls or ceilings, physical facilities do not appear to be vitally important in school choice, particularly for parents. What does appear to matter more is a less tangible quality relating to governance and management of schools, in turn related to head teacher leadership. Parents appear to recognise a difference in schools in which they are more involved, both in their children‟s learning, but also in the policies and practices of the school.

Some government schools are criticised because of a lack of accountability15

among

teachers, whom parents have little power to incentivise to be present or to perform better, and who are seen to treat teaching as a sinecure (it should be stressed that not all government schools or teachers are seen in this light). The quantitative data on governance fail to capture these significant differences between newer and unregistered institutions compared to the government and registered schools. This is partly because on the quantifiable indicators of school governance (SMC presence and activity, PTAs, official school inspections, etc), registered and government schools perform significantly better than other types of school. A fuller and more meaningful assessment of education governance in Bangladesh will depend on developing indicators that reflect how governance and management operate in practice; in turn, this will require an understanding of the differences in governance and management between the new and unregistered institutions compared to the established government and registered schools, as this appears to be an important area of difference in how these are judged.

A final important finding with respect to school supply is that this varies considerably by area. In addition, while there are many more schools available than was the case 20 years ago, in practice, choice remains constrained by the two most important determinants of school choice: cost and access. While the constraints of cost in terms of capacity to invest in education are well-established, it is far less widely registered that physical access continues to present a significant constraint. An important reason for this is time and financial cost, but also concerns about children‟s, and particularly adolescent girls‟ security. The present study found evidence to support the view that security concerns are significant factors limiting school choice.

6.2 Preferences and aspirations

In terms of preferences for different types and levels of education and aspirations for children‟s schooling, the study found that a basic minimum level of education was the norm. The level at which this minimum was set varied between the areas, and appeared to be affected both by the economic opportunities available for which education might be required, as well as by the social norms within the relevant reference group, or community. Up to primary, there appeared to be no discernible gender differences in aspirations for children‟s education; at secondary, differences emerged, although these were not all in favour of boys‟ education. In areas where international migration was an option, a minimum level of education for boys that was adequate to gain entry into that labour market was deemed desirable; by contrast in some communities, girls‟ education was seen as a means of enhancing marriage prospects. Madrassah and/or Arabic education were seen as particularly attractive for girls, and in view of their marriage potential.

15

The Bangla word for accountability, jobabdihita was, specifically, used in this context.

55

Area effects were also strong with respect to the impact of local livelihood opportunities on aspirations for education. Where low-skilled agricultural wage labour was the currently most likely future livelihood, basic ability to read, write and keep accounts was considered necessary. By contrast, in the most central area, there were strong preferences for investment in education to higher levels, seen as having potential for social mobility and formal sector employment. Aspirations were not entirely determined by local livelihood opportunities, however, reflecting the possibility of mobility for employment and growing awareness of opportunities elsewhere in and beyond Bangladesh.

The desire for social status shapes school preferences, and the social status considerations of local reference groups appears to have a significant impact on aspirations, again highlighting the importance of area effects. The crucial role of schools in creating social status is also highlighted here: the widening of school choice appears to have been driven in part by the need for affluent or elite groups to mark distinctions between themselves and poorer households, as these have enrolled in government primary schools in the masses.

An important finding about the perception of quality in schools is that many parents closely scrutinise and possess standards for teaching against which they judge school performance. It does not appear to be the case that most parents are viewing school performance from a position of ignorance or apathy; however, it is not clear that parents have many opportunities or mechanisms through which to exact better performance when they feel teaching is falling short.

In terms of the apparently growing preference for religious schooling, this appears to reflect a number of factors. The present study was limited in the extent to which it could explore the diversity and the content of education service delivery being provided by these institutions, partly because in the absence of pre-existing empirical evidence on these, the study instruments could not be designed to explore all the dimensions of their activity and importance within society. However, a number of findings are worth noting, and some merit further exploration. The first of these is the reasonably well-established fact that the supply of madrassah institutions of a wide range and type has grown in recent years. This partly reflects the growth of public financial support to these institutions (Asadullah and Chowdhury; Bano). In addition, many private madrassahs in our sample were benefiting from the flow of private donations, often from expatriate Bangladeshis in the Middle East, but also from local donations, which have customarily supported local madrassahs and maktabs.

The growth in supply of madrassah institutions appears to reflect in turn a latent demand for religious education. Here the evidence is necessarily more fragmented. However, a number of factors stand out. First, the change in their official recognition and status means that graduates of registered madrassahs now officially stand equal chances of entering general universities and the civil service. This means that some disincentives for attending ebtedayee, dakhil, alim or fazil madrassahs within the officially registered and supported aliya stream have disappeared. Second, the growth of the madrassah sub-sector in itself has entailed a rising demand for madrassah teachers. This was most marked in our Chittagong site, where the largest number of formal and semi-formal madrassah institutions was found. A third factor appears to be the impact of migration to the Middle East, which appears to create a preference for religious education for the children of migrants, possibly in part because it marks a form of social distinction, but also because of beliefs that having Arabic language skills may increase chances of successful overseas migration. The study findings on these points are fragmentary and anecdotal, however, and do not amount to robust evidence in support of the view that migration and religious education are closely related because of an assumption of future job prospects. A fourth factor on which ample qualitative evidence was gathered is that in many households, placing at least one child in a madrassah

56

is seen as a source of religious blessings and social respect; in a number of accounts, children were enrolled in madrassahs to fulfil the dying wish of a family member, or to honour a religious promise or prayer.

6.3 Actual schooling choices and outcomes

Measured in terms of the levels of education attained and diversity of institutions attended, the educational opportunities of the present generation of school going children are considerably wider and deeper than was the case for their parents. They are much more likely to enrol and to stay in school and to attend a wider range of types of school, particularly at primary, than their parents. However, over two-thirds in our sample still attend government primary schools, highlighting the significant limits to choice in practice. No particularly significant gender differences could be identified in schooling choices, which is in itself a significant finding.

No particular patterns with respect to school choice could be identified by poverty level, in addition, although this may reflect the nature of the sampling strategy and the small sample size as much as any real lack of difference. However, area effects appear to be strong, most notably in the levels of further and higher education households aim at for their children, which are the intermediate outcomes of basic education for some.

The study was limited in its capacity to explore the socioeconomic outcomes of schooling choice decisions, given the effects of area on school choice, the predominance of the public sector in provision and the relatively recent phenomenon of the expansion of school choice. Some findings included that level of education attained was most important for women‟s labour force participation, and that expectations of educational attainment varied by area, in support of other findings of the study. The majority of men household members in our sample who were no longer in education reported their activities as self-employment, which covers a range of activities from formal sector businesses, through to petty trading. This seems to contrast with previous occupational pattern, in which most rural adult men could be expected to report their occupation as landowners, farmers or agricultural wage labourers, and seems to highlight the growth of off-farm and non-farm economic activity in rural areas. Qualitative research suggested that migration and the madrassah sector itself could boost demand for religious schooling, although this was balanced by the view that formal sector positions, particularly in government, require general stream education.

The study also attempted to explore outcomes in terms of socio-political attitudes of students emerging from madrassah compared to general stream schools. The findings are suggestive rather than definitive, of some emergent differences in attitudes. These included an overall strong preference for democratically elected leadership, for which madrassah students showed only very slightly less inclination than non-madrassah students. Across the body of student respondents, a preference for democratic leadership appeared to be compatible with agreement with rule by experts (the mode of rule at the time of the survey) as well as significant support for military rule. The only significant differences between madrassah and non-madrassah students emerged in relation to attitudes towards rule by inherited and by religious leaders, of which madrassah students were considerably more supportive. With respect to beliefs about democracy, no important differences across school type were found, although one concerning finding was the overall lack of trust in democratic governance demonstrated by the widely held view that corruption increases in democracy. On citizen rights, the significant finding was the wide differences in attitudes to women‟s freedom to work: while nearly two-thirds of boys in madrassah institutions who responded disagreed with this freedom, 86 per cent of girls in non-madrassah institutions who responded agreed with it, indicating a wide area of disagreement on a fundamental principle of women‟s rights.

57

6.4 Policy implications

The School Choice in Bangladesh study has documented social changes being brought about by changing attitudes and preferences, policies and market forces in basic education in Bangladesh. It was based on four purposively selected community case studies, and therefore the findings cannot be assumed to be representative of trends across the country. However, the homogeneity and centralisation of the basic education sector mean that other communities are indeed likely to be experiencing many of the processes of change discussed here. However, the study also demonstrated the impacts both of economic livelihood opportunities in shaping educational choice, but also of area effects, including locally formed preferences for culture and learning.

One broad-based and generalisable issue of interest is the role of ease of access in influencing choice, and the finding here that concerns about children‟s safety are a significant element in school decisions. This finding has policy implications beyond the specific issue of school choice, with bearings on the broader question of access for the hard-to-reach or the out-of-school children. It also highlights that issues of physical access have not been resolved in the basic education sector. A second issue on which some generalisation can be made is the significance of accountability and participation in shaping school preferences. People do appear to value certain forms of interaction and school governance found in the non-registered schools. However, the tools commonly used to measure and assess how these shape the school experience were inadequate when it came to analysing the different accountability relationships between parents and madrassahs or fee-paying KG or even NGO schools.

A key conclusion to be considered with respect to policy development includes that wider choice may not lead to significant gains in terms of quality. Policies of supporting wider school choice on grounds that this will lead to improvements in quality may not yield the desired outcomes, partly, but not entirely, because of real constraints of cost and access. Voucher schemes and stipend programmes which do not significantly address parents‟ ability to engage with teachers and to effectively exact higher standards of performance are unlikely to produce better education services. And the effective choice set is in any case likely to be narrow in most rural communities, particularly at primary.

Differences of quality appear to reflect differences of governance and management at least as much as tangible indicators of physical facilities or teacher qualifications. One issue here is that commonly used indicators of governance and management do not accurately capture some of the dimensions of school performance valued by parents. The roles of madrassahs in hosting Awaaz Mahfil and other widely-attended religious events, and the relationship between the fee-paying parent and the salary-dependent KG school teacher differ greatly from how participation and accountability are supposed to take place in the registered and government schools. Meaningful indicators of governance and management or accountability and participation will need to be developed in conjunction with a full understanding of how schools within the government system compare with the newer and unregistered entrants the education system.

The study suggests reasons to be concerned about middle class flight, although it was not equipped to prove this was taking place on scale. Although for a number of methodological reasons, the study did not set out to trace households which had left the government system for private schools, there are good reasons to believe that there has been a move towards private KG schools, where these are available and for those who can afford the investment. If the indicative findings of this study are replicated on a wider scale, the government registered sector is at risk of becoming schools where only poorer children attend because

58

they cannot afford to choose to go elsewhere. The implications of this include a likely further deterioration of quality, as pressures on teachers from educated parents are removed, and over the longer term, a declining political priority for public education policy and financing.

A fuller understanding of the diversity and complexity of the religious education sub-sectors and the phenomenon of the apparently growing preference for religious education requires further empirical work and analysis. The experience of the present study was that such work should be feasible, and that entry into madrassahs to conduct surveys, including quomi, hafezia, noorani and forkania madrassahs, is possible. The research team‟s experience was that having madrassah-educated researchers in the team eased access, but also that

resistance to outsiders and researchers was not as strong as had been anticipated.16

A related issue is how to evaluate choices of religious education, and understand what they mean to poor households. Recent evidence from CAMPE (2008) supports the widely-held view that even within the registered system, investments in and returns from madrassah education are relatively low. This in turn supports a tentative concern about potential future grievances arising from unequal support for and unequal outcomes from madrassah education. The research found no reason to believe that any such grievance is currently perceived or being politicised. It helps also to understand what the choice of religious education means for poor families. For many, the ability to choose religious schooling is very likely a mark of social progress that is both affordable and meaningful; it may therefore not signal a growing reversion to „traditional‟ cultural or religious practices but is about realistic aspirations.

The growth of mass private education sub-sectors, particularly at primary, is clearly a matter of considerable importance, with potential implications for the governance, management, performance, and above all student intake into the government and government-supported system. While many issues are core policy matters of curriculum, recognition status, and inspection and supervision, there are also some important issues for analysis over the longer-term. Some issues that may merit further consideration include the matter of private financing of education. This includes the religious sub-sector, whose private charitable financial flows remain an interesting unknown area. In the interests of increasing school choice and quality, however, policymakers may consider the growing significance of other private financial flows, as well as the potential significance of private charity to education more widely. There appears to be considerable but not quantified support for poorer students to attend school, which warrants further attention from a number of education policy perspectives, including equity, sustainability of education financing, and community participation in and management of education.

16

The one quomi madrassah in the sample had in the previous week been visited by the authorities, reportedly seeking a terrorist associated with an Islamic extremist group, whom they had failed to find evidence of on-site. Yet they were welcoming of the male researchers in the team, and gave a full interview.

59

References

Abdalla, A., Raisuddin, A. N. M., and Hussein, S., 2004, „Bangladesh Educational Assessment: Pre-primary and Primary Madrasah Education in Bangladesh‟, Basic Education and Policy Support (BEPS) Activity/United States Agency for International Development Asadullah, M.N. and Chaudhury, N., 2006 „Religious Schools, Social Values and Economic Attitudes: Evidence from Bangladesh‟, QEH working paper series.

Al-Samarrai, S., 2006, „Where have all the boys gone? A note on the recently released PEDP II baseline enrolment statistics‟, Dhaka: unpublished mimeo

Al-Samarrai, S., 2008a, „Changes in educational attainment in Bangladesh, 2000-2005‟, background paper for the World Bank poverty assessment, Dhaka

Al-Samarrai, S., 2008b, „Education and Labour force outcomes‟, background paper for World Bank poverty assessment, Dhaka

Amin, S.; Selim, N. and Kamal Waiz, N., 2006, Causes and Consequences of Early Marriage in Bangladesh, background report for Workshop on Programmes and Policies to Prevent Early Marriage, Dhaka: Population Council

Bano, M., 2007, „Allowing for Diversity: State-Madrasa relations in Bangladesh‟, Religions and Development Research Programme

Bangladesh Bureau of Educational Information and Statistics (BANBEIS), 2005, Retrieved from http://www.banbeis.gov.bd/.

BRAC Annual Report 2007, Dhaka: BRAC

(http://www.brac.net/downloads_files/BRAC_Annual_Report_2007.pdf)

Chakrabarti, R., 2007 „Can increasing private school participation and monetary loss in a voucher program affect public school performance? Evidence from Milwaukee‟, staff reports, Federal Reserve Bank of New York

Directorate of Primary Education (DPE), 2006a, „Baseline survey of Second Primary Education Development Programme (PEDP-II) - draft report‟. Dhaka, Ministry of Primary and Mass Education, Government of the Peoples' Republic of Bangladesh: 92.

Evans, G. and Rose, P., 2007, “Testing Mechanisms of Influence: Education and Support for Democracy in Sub-Saharan Africa.” Afrobarometer series working paper no. 75.

FMRP, 2006, „Social Sector Performance Surveys: Primary education in Bangladesh: Assessing service delivery‟. Dhaka, OPM and FMRP.

FMRP, 2006, „Social Sector Performance Surveys: Secondary education in Bangladesh: Assessing service delivery‟. Dhaka, OPM and FMRP.

Household Income and Expenditure survey, 2001, Bangladesh Bureau of Statistics, Dhaka.

Hossain, N. and Kabeer, N., 2004, „Achieving universal primary education and eliminating gender disparity,‟ Economic and Political Weekly, vol. 39, No 36 Sep 4-10.

http://www.banbeis.gov.bd/

http://www.brac.net/downloads_files/BRAC_Annual_Report_2007.pdf

60

Hoxby, C. M., 2003, „School choice and school productivity: could school choice be a tide that lifts all boats?‟ Hoxby, C. M. (ed.), The Economics of School Choice. Chicago, Ill., University of Chicago Press, pp. 287–342. (A National Bureau of Economic Research Conference Report.)

Kabeer, N., 2001, 'Intergenerational contracts, demographic transitions and the 'quantity-quality' trade-off: children, parents and investing in the future', Journal of International Development Vol 12(4): 463-482

Levin, H. M. and Belfield, C. R., 2004, „Vouchers and public policy: when ideology trumps evidence‟, New York: National Center for the Study of Privatization in Education. Columbia University

Rahman, A. et al., 2005, Bangladesh Health and Injury Survey: Report on Children, Directorate General of Health Services, Ministry of Health and Family Welfare, Government of the People's Republic of Bangladesh

Saferworld, 2008, Human Security in Bangladesh, Dhaka: Security in South Asia Report, Saferworld

UNESCO, 2008, Global Monitoring Report 2008, Paris: UNESCO.

61

Annexes

A.1 Tables

Table A.1 Average number of education institutions in rural upazilas¹

National Bhurungumari, Kurigram

Kumarkhali, Kushtia

Kaligonj, Gazipur

Rangunia, Chittagong

GPS 75 62 81 85 93

RNGPS 41 44 50 23 36

NRNGPS 3 0 5 1 2

English Medium 16 21 0 3 1

Kindergarten 1 6 0 2 1 5

Kindergarten 2 (NGO)

2 0 0 0 1

Community 7 2 8 13 9

Attached Ebtedayee madrassahs

18 17 10 28 14

Attached High School

4 0 3 8 0

BRAC primary schools

Secondary 35 32 53 37 40

Madrassahs 19 17 20 30 14

Notes: 1) Metropolitan and shadar upazilas are excluded. 2) Source: Columns 1-9: Directorate of Primary

Education, PEDP II baseline survey 2005, Column 10: BRAC Education Programme data, Columns 11-12: BANBEIS data for 2005.

62

Table A.2 Average fees and other school charges per annum by school type

and area (Taka)

Educational Institutions Kurigram Gazipur Kushtia Chittagong

Government primary school 34 80 52 39

Non-govt primary school 35 364 115

NGO primary school 36

Community school 241 52 40

Kindergarten/English medium 218 393 45 80

Ebtedayee madrassah 210 440 233 47

Quomi madrassah 355 327 109

Unreg. non-govt primary school

319 628 389 117

Govt secondary school 540 567 116

Non-govt secondary school 298 605 42

Non-govt junior secondary 1350 243

Non-govt technical college 264 564 312

English medium secondary school

495 390 156

Cadet college 560 915 231

Coaching centre 438 521

NGO school (secondary) 67

Dakhil madrassah 607 49

Quomi madrassah 34 459

Unreg. non-govt secondary 40 208

Govt intermediate college 38 471

Non-govt intermediate college 278

Govt degree/masters college 557 223

Non-govt degree/masters college

525 511

Alim madrassah 1313 223

Fazil madrassah 265 494

Kamal madrassah 687 1700

Other 252 850


63

Table A.3 Average annual household spending by poverty ranking and school type (Taka)

Educational Institutions

1st Quartile 2nd Quartile 3rd Quartile 4th Quartile

Primary Institutions:

Government primary school 1809.17 2302.50 2409.75 3980.00


- 2329.00 - 1400.00

NGO primary school 1855.00 1149.00 1825.00 -

Community primary school 445.00 - - -

Kindergarten 815.00 - 6880.00 -

Primary madrassah - - 3899.00 800.00

Secondary Institutions:


1145.00 3237.00 3294.00 7621.56


1305.00 400.00 6402.67 17577.00

Dakhil madrassah - 1532.67 1903.75 1998.00

Secondary Quomi madrassah

- 4070.00 - -

Higher Institutions:

Alim madrassah 7798.00 1100.00 - 10500.00

Fazil madrassah - - 2486.67

Non-formal madrassahs:

non-formal madrassah - 6590.00 - 1595.00

Notes: 1) Based on reporting about one child in the household, through whom the household was sampled (attending a catchment area school). 2) Source: household survey, School Choice study.

64

Table A.4 Male employment outcomes by education level (%)

Employment types Never enrolled or no class completed

Less than 5 classes completed

Classes 1 to 5 completed



Greater than 12 classes completed

Formal sector employment

1% 8% 8% 14% 40% 50%

Informal sector employment

0% 10% 6% 9% 20% 0%

Daily wage labourer 28% 33% 22% 5% 0% 0%

Self-employment 39% 25% 39% 55% 40% 0%

Unpaid domestic work for own household

6% 8% 5% 0% 0% 0%

No current occupation 14% 15% 6% 5% 0% 50%

Other 11% 3% 13% 14% 0% 0%

Total 100% 100% 100% 100% 100% 100%

Notes: 1) source: household survey, School Choice study

65

Table A.5 Female employment outcomes by education level (%)

Employment types

Never enrolled or no class completed

Less than 5 classes completed




Formal sector employment

0% 0% 0% 7% 50%

Informal sector employment

0% 0% 0% 0% 0%

Daily wage labourer 0% 0% 1% 0% 0%

Self-employment 2% 4% 0% 0% 0%

Unpaid domestic work for own household

86% 74% 91% 93% 50%

No current occupation 3% 22% 8% 0% 0%

Other 10% 0% 0% 0% 0%

Total 100% 100% 100% 100% 100%

Notes: 1) source: household survey, School Choice study

66

A.2 The study sites

Summary information on the community case study sites is provided in Table 2.1. More detailed findings on each will be provided in the substantive sections of the report. In brief, the sites can be ranked as follows.

The poorest area is the Kurigram site, with the fewest and least diverse set of schools. Economic opportunities are limited, even for the wealthier households, to working on or managing agricultural land. The land itself is of uneven quality, and highly vulnerable to riverbanks bursting and annual floods. For most households, the prospect of planning investments in education for the future generation is unrealistic, because of the insecurities associated with land erosion and regular environmental crisis. The view is that education fills children‟s time until they are old enough to work to contribute to the household or manage the land (boys) or to marry (girls). Very little community support for educational institutions is forthcoming, and madrassahs have had to close. Age of marriage is low, including for boys in poorer households, because of the dowries they bring. Formal education provision is limited, few people can afford the time or money to send their children to more distant schools, and the quality of instruction in the local government primary school is widely accepted to be extremely poor. BRAC schools are popular, but selective and limited in intake.

The Kushtia site is better off, and there are fewer extreme poor households. The village is dominated by a few very large landlord families, but smallholders are common and most households rely on agricultural production. The area has a rich cultural heritage associated with learning, and a stronger overall preference for schooling than in Kurigram. Although there were fewer institutions within the defined catchment area than in Kurigram, in practice, there is a more meaningful choice of schools, particularly at primary. Security has improved over the last few years, but parents remain reluctant to allow children to travel to schools: a secondary school going girl from the village was kidnapped and raped on her journey home, in a relatively recent incident which caused many parents to rethink their daughters‟ schooling. Compared to the government schools, BRAC and private schools are both considered better performers, although private schools are unaffordable for many, and there are none (other than government-registered institutions) in the village.

In terms of absolute numbers of institutions, as well as households‟ economic capacity to exercise choice, the Gazipur and Chittagong sites are similar. Both feature diverse and numerous institutions, and the majority of the community members are relatively affluent. Other similarities include a high proportion of migrants, and proximity to major cities. The Gazipur site economy was less reliant on agriculture, and featured more formal sector employment, while in Chittagong, agriculture, business and migration were the key activities. The Gazipur community seemed overall to have a greater interest in formal educational investment, and to higher levels, whereas in Chittagong, religious education retains a strong appeal. Both areas feature a highly diverse set of schools, including KG (or kindergarten) and NGO schools, as well as one of the highly elite government secondary schools (in Chittagong), and a Church mission school and a highly-regarded high school attached to an established industry (both in Gazipur).

69

A 3 Participatory school mapping and timeline exercises

Figure A1 presents graphs illustrating how the supply of schools had increased and diversified recently. These graphs were produced through a participatory process designed to explore how school supply had changed over the last generation, undertaken with children, youth and other community members. The institutions identified through this process were not identical with those identified as serving the community under the definition of catchment area and the school listing exercises used in the present study. The children, youth and other community members who participated in the school mapping and timeline exercises mainly identified institutions that fell within the geographical boundaries of the village – not all of which necessarily fell within the catchment area definition, or which local children were generally known to attend. This included some categories of institutions which the present study did not include, such as private tutors, pre-primary and post-basic education institutions.

According to discussions that emerged through the timeline process, all four areas have seen growth in sheer numbers and in diversity since 1990. In Kurigram and Kushtia, growth in numbers and diversity appears to have been less dramatic than in Gazipur and Chittagong. In both Kurigram and Kushtia, people felt that the NGO schools were the newer entrants, and that government school supply had remained static. The numbers and diversity of institutions in and around the Gazipur and Chittagong communities were believed to have grown considerably over the last nearly two decades. In Chittagong, people felt that there had always been some well-established madrassahs in the area, and that coaching centres, KG schools, smaller madrassahs and NGO schools were all relatively arrivals. People in Gazipur noted that their first (now government) primary school was established in 1914, and the high school came soon after. The dakhil madrassah had been there since the 1960s, but among the madrassahs, the maktab was believed to be the oldest, as „the teachers there said they saw it when they were little‟. The two colleges were established in the 1990s. The kindergarten schools and NGO schools arrived around the same time, between 2002 and 2007, and the hafeziya madrassahs were also established around five or six years ago.

71

Figure A1 Participatory timelines showing establishment of schools

Kurigram

Kushtia

0

5

10

15

20

25

30

1930 1940 - 60 1970 - 90 2000 - 2008

2

9

20

30

NO

. OF

SC

HO

OLS

72

Gazipur

Chittagong

0

20

40

60

1970 1980 1990 2008

210

18

51

NO

. OF

SCH

OO

LS

Notes: 1) source: participatory school mapping and time line exercise, School Choice study.

73

FIGURE A2 Participatory school maps

Participatory school map of the Kurigram community

Primary school 4

Secondary School 2

Small Secondary School 1

Madrasa 2

Degree College 2

NGO School (BRAC) 4

Islamic Foundation 2

Mosque 2

Best Primary School

Best Secondary School

Most popular Primary School

74

Most popular Secondary School

76

Participatory school map of the Kushtia community

77

Key:

Primary schools: 8, 10, 21, 23

Secondary schools: 6, 7, 24, 25, 26

NGO schools: 1, 2

Madrassahs and maktabs: 3, 20, 12, 14, 15, 22

Kindergarten: 11, 9

Coaching centres: 4, 5, 27

Colleges: 13

Private tuition centres: 16, 17, 18, 19

78

SYMBOL TYPE

Primary School

Secondary School

Kindergarten

Madrasa

Hafiziya Madrasa

Mosque

Madrasa and Mosque

NGO School (BRAC)

NGO School (World Vision)

Cadet Academy

Missionary School ( Saint

Nicholas Uccho Vidyalaya)

College

79

Participatory school map of Gazipur community

Best Primary School

Best Secondary School

Most Popular Primary School

Most Popular Secondary school

81

Participatory school map of the Chittagong community

82

Key:

Primary schools: 1, 3, 4, 5,6

Secondary schools: 2, 34, 36, 37, 42

Colleges: 33, 35, 38, 48, 49

Technical colleges: 40, 7

NGO schools: 8

Kindergarten (KG) schools: 23, 24

Hindu and Christian schools: 19, 43

Coaching centres: 21, 22, 25, 29, 44, 45, 46, 47

University: 50

Madrasa: 17, 16, 18, 41, 51, 10, 9, 11, 12, 14, 15, 20, 30, 31, 32, 13, 26

SCHOOL CHOICE IN BANGLADESH - Institute of … · The motivation for the School Choice in...

Documents

Transcript of SCHOOL CHOICE IN BANGLADESH - Institute of … · The motivation for the School Choice in...